<?xml version="1.0" encoding="UTF-8"?>
  <?xml-stylesheet type="text/xsl" href="rfc2629.xslt" ?>
  <!-- generated by https://github.com/cabo/kramdown-rfc version 1.7.31 (Ruby 3.3.8) -->


<!DOCTYPE rfc  [
  <!ENTITY nbsp    "&#160;">
  <!ENTITY zwsp   "&#8203;">
  <!ENTITY nbhy   "&#8209;">
  <!ENTITY wj     "&#8288;">

]>

<?rfc strict="yes"?>
<?rfc comments="yes"?>
<?rfc docmapping="yes"?>

<rfc ipr="trust200902" docName="draft-ietf-pquip-pqc-hsm-constrained-12" category="info" consensus="true" submissionType="IETF" tocInclude="true" sortRefs="true" symRefs="true">
  <front>
    <title abbrev="Adapting Constrained Devices for PQC">Adapting Constrained Devices for Post-Quantum Cryptography</title>

    <author fullname="Tirumaleswar Reddy">
      <organization>Nokia</organization>
      <address>
        <postal>
          <city>Bangalore</city>
          <region>Karnataka</region>
          <country>India</country>
        </postal>
        <email>k.tirumaleswar_reddy@nokia.com</email>
      </address>
    </author>
    <author fullname="Dan Wing">
      <organization abbrev="Citrix">Citrix</organization>
      <address>
        <postal>
          <country>United States of America</country>
        </postal>
        <email>danwing@gmail.com</email>
      </address>
    </author>
    <author fullname="Ben Salter">
      <organization>UK National Cyber Security Centre</organization>
      <address>
        <email>ben.s3@ncsc.gov.uk</email>
      </address>
    </author>
    <author fullname="Kris Kwiatkowski">
      <organization>PQShield</organization>
      <address>
        <email>kris@amongbytes.com</email>
      </address>
    </author>

    <date year="2026" month="November" day="15"/>

    <area>Security</area>
    <workgroup>PQUIP</workgroup>
    <keyword>PQC</keyword> <keyword>IoT</keyword> <keyword>TEE</keyword> <keyword>HSM</keyword> <keyword>RoT</keyword>

    <abstract>


<?line 183?>

<t>This document provides guidance on integrating Post-Quantum Cryptography (PQC) into
resource-constrained devices, such as IoT nodes and dedicated key-storage hardware.
These systems often operate with strict limitations on processing power, RAM, and
flash memory, and may even be battery-powered. The document emphasizes the role of hardware
security as the basis for secure operations, supporting features such as seed-based key
generation to minimize persistent storage, efficient handling of ephemeral keys, and the
offloading of cryptographic tasks in low-resource environments. It also explores the
implications of PQC on firmware update mechanisms in such constrained systems.</t>



    </abstract>

    <note title="About This Document" removeInRFC="true">
      <t>
        Status information for this document may be found at <eref target="https://datatracker.ietf.org/doc/draft-ietf-pquip-pqc-hsm-constrained/"/>.
      </t>
      <t>
        Discussion of this document takes place on the
        pquip Working Group mailing list (<eref target="mailto:pqc@ietf.org"/>),
        which is archived at <eref target="https://mailarchive.ietf.org/arch/browse/pqc/"/>.
        Subscribe at <eref target="https://www.ietf.org/mailman/listinfo/pqc/"/>.
      </t>
    </note>


  </front>

  <middle>


<?line 194?>

<section anchor="introduction"><name>Introduction</name>

<t>The transition to post-quantum cryptography (PQC) poses significant challenges for
resource-constrained devices, such as dedicated key-storage hardware and
Internet of Things (IoT) devices.</t>

<t>These devices typically operate under strict limitations on
processing power, RAM, and flash memory, and in some cases are battery-powered. Adopting
PQC algorithms in such environments is difficult due to their substantially larger key
sizes and, in some cases, higher computational demands. Consequently, the migration to
PQC requires careful planning to ensure secure and efficient key management within
constrained platforms.</t>

<t>This document does not specify protocol behaviour or algorithm selection for constrained networks and leaves this to the relevant protocol specifications. Some PQC schemes covered in this document have public keys, ciphertexts or signatures one to two orders of magnitude larger than their elliptic-curve counterparts (see <xref target="sec-key-sizes"/>). On links with small frames, such as IEEE 802.15.4, sending those artifacts requires fragmentation across many frames, which increases latency, energy consumption and exposure to loss, particularly in multi-hop mesh networks.</t>

<t>Constrained devices are often deployed as clients initiating outbound connections, but some also act in server roles or enforce local authentication policies.
As a result, designers may need to consider PQC to address confidentiality, both outbound and inbound authentication, and signature verification used in secure boot, firmware updates, and device attestation.</t>

<t>In this document, "constrained device" is used as an umbrella term for two related cases:</t>

<t><list style="symbols">
  <t>constrained nodes, as defined in <xref target="I-D.ietf-iotops-7228bis"/>. This document primarily targets Class 1 devices and above. Class 0 devices generally lack the RAM and storage needed for the algorithms covered here, and would typically need to offload PQC operations to a dedicated cryptographic module.</t>
  <t>constrained cryptographic modules with limited RAM, storage, or processing capacity, which may be embedded in hosts that are not themselves constrained.</t>
</list></t>

<t><xref target="key-management"/> applies mainly to constrained cryptographic modules, while <xref target="sec-key-sizes"/>, <xref target="additional-considerations"/>, <xref target="sig-mem"/>, <xref target="sig-perf"/> and <xref target="firmware"/> apply to both constrained nodes and modules as a whole.</t>

<t>This document provides guidance and best practices for integrating PQC algorithms into constrained devices. It reviews strategies for key storage, ephemeral key management, and performance optimization tailored to low-resource environments. The document also examines ephemeral key generation in protocols such as TLS, along with techniques to optimize PQC signature operations within constrained cryptographic modules. It is intended for implementers of the constrained devices and cryptographic modules described above. It is also intended for vendors that integrate such modules from third parties, to help them determine which capabilities and optimizations to require from their suppliers.</t>

<t>For new designs, PQC requirements should be an input to hardware selection rather than being fitted afterwards to hardware chosen for other reasons. The larger keys, signatures and working memory of PQC algorithms (see <xref target="sec-key-sizes"/> and <xref target="sig-mem"/>), together with the need to accommodate potential parameter changes over the device lifetime, should be accounted for when sizing RAM, storage and processing capacity. The techniques in this document help where the hardware is already fixed, such as for existing deployments or third-party modules.</t>

<t>The memory optimization techniques in <xref target="sig-mem"/> are based on published implementation results on ARM Cortex-M class microcontrollers. The analysis of ML-DSA rejection sampling follows <xref target="KWI2026"/>, with an open-source reference implementation available in <xref target="KWI2026CODE"/>. Actual feasibility depends on the target platform and should be validated on it.</t>

<t>This document focuses on PQC algorithms standardized by the US National Institute of Standards and Technology (NIST) or specified by the IRTF CFRG, and that have corresponding IETF protocol specifications, either published as RFCs or progressing through the IETF standards process. Specifically, it covers the following algorithms:</t>

<t><list style="symbols">
  <t>Module-Lattice-Based Key-Encapsulation Mechanism (ML-KEM) <xref target="FIPS203"/>.</t>
  <t>Module-Lattice-Based Digital Signature Algorithm (ML-DSA) <xref target="FIPS204"/>.</t>
  <t>Stateless Hash-Based Digital Signature Algorithm (SLH-DSA) <xref target="FIPS205"/>.</t>
  <t>Hierarchical Signature System/Leighton-Micali Signature (HSS/LMS) <xref target="RFC8554"/>, and the related eXtended Merkle Signature Scheme (XMSS) <xref target="RFC8391"/>.</t>
</list></t>

<t>Additional post-quantum algorithms are expected to be standardized in future, which may also prove suitable for use in constrained devices. Since algorithms may change prior to standardisation (or may end up unstandardized), no concrete guidance is provided on these here, but future specifications may provide guidance on the following algorithms:</t>

<t><list style="symbols">
  <t>The Falcon signature scheme <xref target="Falcon"/> has shorter keys and signatures than ML-DSA, which makes it appealing for constrained devices. However, its reliance on Gaussian sampling makes it difficult to implement correctly and to validate against implementation attacks. A revised specification is awaited at the time of writing.</t>
  <t>The HQC <xref target="HQC"/> is a code-based KEM, so offers algorithmic diversity to complement lattice-based KEMs, though it is expected to be less performant than ML-KEM.</t>
  <t>Smaller SLH-DSA parameter sets <xref target="Smaller-SPHINCS"/> may be standardized in future, which may make use of SLH-DSA more palatable on constrained devices.</t>
</list></t>

<t>This document focuses on device-level adaptations and considerations necessary to implement PQC efficiently on constrained devices.
Actual protocol behaviour is defined in other documents.</t>

</section>
<section anchor="key-management"><name>Key Management in Constrained Devices for PQC</name>

<t>The embedded cryptographic components used in constrained devices are designed to securely manage cryptographic keys, often under strict limitations in RAM, flash memory, and computational resources. These limitations are further exacerbated by the increased key sizes and computational demands of PQC algorithms.</t>

<t>One mitigation of storage limitations is to store only the seed rather than the full
expanded private key, as the seed is far smaller and can derive the expanded private key
as necessary. <xref target="FIPS204"/> Section 3.6.3 specifies that the seed ξ generated during ML-DSA.KeyGen can be stored for later use with ML-DSA.KeyGen_internal.
To reduce storage requirements on constrained devices, private keys for
Initial Device Identifiers (IDevIDs) and Locally Significant Device
Identifiers (LDevIDs) <xref target="IEEE802.1AR"/>, and the optional attestation private key can be
stored as seeds instead of expanded key material.</t>

<section anchor="Seed"><name>Seed Management</name>

<t>The following is some additional guidance to aid in compliance with <xref target="FIPS203"/>, <xref target="FIPS204"/>, <xref target="FIPS205"/> and <xref target="REC-KEM"/>:</t>

<section anchor="seed-storage"><name>Seed Storage</name>

<t>Several post-quantum algorithms use a seed to generate their private keys (e.g., ML-KEM and ML-DSA). Those seeds are smaller than private keys, hence some implementations may choose to retain the seed rather than the full private key to save on storage space. The private key can then be derived from the seed when needed or retained in a cache within the security module.</t>

<t>The seed is a Critical Security Parameter (CSP) as defined in <xref target="ISO19790"/>, from which the private key can be derived, hence it must be safeguarded with the same
level of protection as a private key. Seeds should be securely stored within a cryptographic module of the device whether hardware or software-based to protect against unauthorized access.</t>

<t>The choice between storing a seed or an expanded private key involves trade-offs
between storage efficiency and performance. Some constrained cryptographic modules may
store only the seed and derive the expanded private key on demand, whereas others may
prefer storing the full expanded key to reduce computational overhead during key usage.</t>

<t>The choice between storing the seed or the expanded private key has direct
implications on performance, as key derivation incurs additional computation. The impact
of this overhead varies depending on the algorithm. For instance, ML-DSA key generation,
which primarily involves polynomial operations using the Number Theoretic Transform (NTT) (see Section 7.5 of <xref target="FIPS204"/>)
and hashing, is computationally efficient compared to other post-quantum schemes. In contrast,
SLH-DSA key generation requires constructing a Merkle tree and multiple Winternitz One-Time
Signature (WOTS+) key generations (see Section 5 of <xref target="FIPS205"/>), making it significantly more computationally intensive. In
many embedded deployments, SLH-DSA is expected to be used primarily for firmware verification.
In this case the device holds only the SLH-DSA public key; the corresponding private key is known
solely to the firmware signer, and key generation is performed on the signer's infrastructure
rather than on the device. Consequently, SLH-DSA key generation cost does not impact device
performance. However, in scenarios where the device generates its own SLH-DSA key pairs, the
higher key generation cost may influence seed-storage design decisions and depend on performance
considerations or standards compliance (e.g., <xref target="PKCS11"/>).</t>

<t>While vulnerabilities like the "Unbindable Kemmy Schmidt" misbinding attack <xref target="BIND"/> demonstrate
the risks of manipulating expanded private keys in environments lacking hardware-backed
protections, these attacks generally assume an adversary has some level of control over
the expanded key format. However, in a hardware-backed protected environment, where private
keys are typically protected from such manipulation, the primary motivation for storing
the seed rather than the expanded key is not directly tied to mitigating such misbinding attacks.</t>

<t>The expanded private key is derived from the seed using a one-way cryptographic function.
As a result, if the seed is not retained at key generation time, it cannot be reconstructed
from the expanded key (as the reverse operation is computationally infeasible). Implementations
should account for this non-recoverability when designing seed management.</t>

<t>A challenge arises when importing an existing private key into a system designed to
store only seeds. When a user attempts to import an already expanded private key, there is
a mismatch between the key format used internally (seed-based) and the expanded private
key. This issue arises because the internal format is designed for efficient key storage
by deriving the private key from the seed, while the expanded private key is already fully
computed. As NIST has not defined a single private key format for PQC algorithms, this
creates a potential gap in interoperability. To address this, the private key formats for
ML-DSA <xref target="RFC9881"/> and ML-KEM <xref target="RFC9935"/> recommend retaining and exporting the seed.
When an imported private key contains both the seed and the expanded key, the
consistency check in Section 8.2 of <xref target="RFC9881"/> or Section 8 of <xref target="RFC9935"/> applies.</t>

</section>
<section anchor="efficient-key-derivation"><name>Efficient Key Derivation</name>

<t>When storing only the seed in a constrained cryptographic module, it is crucial that
the device is capable of deriving the private key efficiently whenever required. However,
repeatedly re-deriving the private key for every
cryptographic operation adds computation to each operation, increasing CPU time and therefore
latency and energy consumption, which can be problematic for battery-powered devices. In scenarios
where these costs are critical, it may be more efficient to store the expanded
private key directly (in addition to the seed). Implementations may choose to
retain (cache) several recently-used or frequently-used private keys to avoid the computational
overhead and delay of deriving private keys from their seeds for each operation.</t>

<t>The key derivation process, such as ML-KEM.KeyGen_internal for ML-KEM or similar
functions for other PQC algorithms, must be implemented in a way that can securely operate
within the resource constraints of the device. If using the seed-only model, the derived
private key should exist only transiently, held for the duration of the cryptographic operation,
and any state derived from it should be securely erased or otherwise made
unrecoverable as soon as it is no longer needed. However, storing the expanded private key may be a
more practical solution in time-sensitive applications or for devices that frequently
perform cryptographic operations.</t>

</section>
<section anchor="exporting-seeds-and-private-keys"><name>Exporting Seeds and Private Keys</name>

<t>Given the potential for hardware failures or the end-of-life of devices containing keys, it
is essential to plan for backup and recovery of cryptographic seeds and private keys.
Constrained devices should support secure seed- or key-backup mechanisms, leveraging protections such as encrypted storage and ensuring that security measures are in place so that the backup data is protected from unauthorized access.</t>

<t>When exporting a seed or private key, the key-encryption key or the key protecting the secure channel used for direct transfer should provide a security strength at least matching the PQ security level of the exported key. Using the security level mapping in <xref target="RFC9958"/>, Level 1 corresponds to AES-128, Level 3 to AES-192, and Level 5 to AES-256; for example, an ML-KEM-1024 or ML-DSA-87 key (Level 5) should be protected using AES-256.</t>

<t>There are two distinct approaches to exporting private keys or seeds from a constrained device:</t>

<section anchor="direct-transfer"><name>Direct Transfer Over a Secure Channel</name>

<t>In scenarios where the constrained device can establish a secure channel to a peer, the device can transfer encrypted private key material directly to another cryptographic module over that channel. The secure channel needs to provide mutual authentication of both endpoints, confidentiality and integrity protection of the transferred material, and end-to-end protection. A mutually authenticated TLS 1.3 <xref target="RFC9846"/> connection is one example of a protocol providing these properties; DTLS 1.3 <xref target="RFC9147"/> offers the same properties over datagram transport and may be more suitable for some constrained deployments.</t>

<t>Since private key material is a long-lived secret, its transfer is particularly exposed to the "harvest now, decrypt later" (HNDL) attack: an attacker records the protected traffic today and decrypts it once a "cryptographically relevant
quantum computer" (CRQC) is available. To mitigate this threat, the secure channel must be established with a key exchange that provides post-quantum security; for (D)TLS 1.3, this can be achieved with a hybrid key exchange combining ECDHE with ML-KEM <xref target="RFC10024"/> or with a standalone ML-KEM key exchange <xref target="I-D.ietf-tls-mlkem"/>. Post-quantum key exchange alone is sufficient to protect against HNDL, as authentication cannot be broken retroactively; however, once CRQCs are available, an attacker could impersonate an endpoint during channel establishment, so post-quantum authentication, e.g., with ML-DSA <xref target="I-D.ietf-tls-mldsa"/>, should additionally be used.</t>

</section>
<section anchor="encrypted-export"><name>Export of Encrypted Seeds and Private Keys</name>

<t>In more common constrained device scenarios for secure exporting of seeds and private keys, a strong symmetric encryption algorithm, such as AES Key Wrap with Padding (<xref target="RFC5649"/>), should be used to encrypt the seed or private key before export. <xref target="RFC5649"/> adds padding to handle key material whose length is not a multiple of 8 octets, such as an expanded private key that does not fall on that boundary.</t>

<t>Operationally, the exported data and the symmetric key used for encryption must both be protected against unauthorized access or modification.</t>

</section>
<section anchor="security-requirements-for-export-operations"><name>Security Requirements for Export Operations</name>

<t>The encryption and decryption of seeds and private keys must occur entirely within the cryptographic modules to reduce the risk of exposure and ensure compliance to established security standards.</t>

</section>
</section>
</section>
<section anchor="ephemeral-key-management"><name>Ephemeral Key Management</name>

<t>Given the increased size of PQC key material, ephemeral key management will have to
be optimized for both security and performance.</t>

<t>For PQC KEMs, ephemeral key pairs are generated from an ephemeral seed, that is used
immediately during key generation and then discarded. Furthermore, once the shared secret is
derived, the ephemeral private key will have to be deleted. Since the private key resides in the
constrained cryptographic module, removing it optimizes memory usage, reducing the footprint of
PQC key material in the cryptographic module. This also ensures that no unnecessary secrets
persist beyond their intended use.</t>

<t>Additionally, ephemeral keys, whether from traditional ECDH or PQC KEM algorithms, are intended
to be unique for each key exchange instance and kept separate across connections (e.g., TLS).
Deleting ephemeral keying material after use helps ensure that key material cannot be reused across connections, which would otherwise introduce security and privacy issues.</t>

<t>Constrained devices implementing PQC ephemeral key management will have to:</t>

<t><list style="symbols">
  <t>Generate ephemeral key pairs on-demand from an ephemeral seed stored temporarily within the cryptographic module.</t>
  <t>Enforce immediate seed erasure after the key pair is generated and the cryptographic operation is completed.</t>
  <t>Delete the private key after the shared secret is derived.</t>
  <t>Prevent key reuse across different algorithm suites or sessions.</t>
</list></t>

</section>
</section>
<section anchor="sig-mem"><name>Optimizing Memory Footprint in Post-Quantum Signature Schemes</name>

<t>A key consideration when deploying post-quantum cryptography in cryptographic modules is the amount and type of memory available. In constrained devices, it is important to distinguish between volatile memory (RAM), used for intermediate computations during cryptographic operations, and non-volatile storage (e.g., flash), used for storing keys, firmware, and configuration data. For instance, ML-DSA, unlike traditional signature schemes such as RSA or ECDSA, requires significant RAM during signing due to multiple Number Theoretic Transform (NTT) operations, matrix expansions, and rejection sampling loops. These steps involve storing large polynomial vectors and intermediate values, making ML-DSA more memory-intensive.</t>

<t>Some constrained systems, particularly battery-operated devices, may have limited RAM available for cryptographic operations, even if sufficient non-volatile storage is available. In such cases, straightforward implementations of PQ schemes may exceed available RAM, making them infeasible without optimization.</t>

<t>Several post-quantum schemes can be optimized to reduce the memory footprint of the algorithm. For instance, SLH-DSA has two flavors: the "f" variants which are parameterized to run as fast as possible, and the "s" variants which produce shorter signatures. Developers wishing to use SLH-DSA may wish to utilize the "s" variants on devices with insufficient RAM to use the "f" variants. Further optimizations may be possible by running the signature algorithm in a "streaming manner" such that constrained device does not need to hold the entire signature in memory at once, as discussed in <xref target="Stream-SPHINCS"/>.</t>

<t>Implementations may trade off resource usage across CPU, RAM, and non-volatile storage. For example, techniques such as lazy expansion reduce RAM usage at the cost of increased computation, while storing expanded key in non-volatile storage can reduce runtime overhead. Designers should balance these trade-offs based on the target platform.</t>

<t>Both the ML-KEM and ML-DSA algorithms were selected for general use. Two optimization techniques that can be applied to make ML-DSA more feasible in constrained cryptographic modules are discussed in <xref target="lazy-expansion"/> and <xref target="pre-hashing"/>.</t>

<section anchor="memory-requirements-of-lattice-based-schemes"><name>Memory requirements of Lattice-Based Schemes</name>

<t>Both ML-KEM and ML-DSA are built on the same lattice structure, and the dominant source of memory usage in either is holding the expanded matrix A and the associated polynomial vectors needed to compute a noisy affine transformation of the form t = A*s + e, where A is a public matrix derived from a seed, and t, s and e are polynomial vectors. In ML-DSA this transformation is written t = A*s1 + s2, with s1 and s2 being the secret polynomial vectors that are part of the private key and are used during signing. The elements of those matrices and vectors are polynomials with 256 integer coefficients modulo Q. For the unpacked representation assumed below, ML-DSA's 23-bit modulus uses 4-byte coefficient words, whereas ML-KEM's 12-bit modulus uses 2-byte coefficient words; packed or otherwise optimized implementations may use different representations. In both schemes the modulus is the same regardless of parametrization. The dimensions of the matrix and vectors, however, do depend on the parameter set.</t>

<t>The worked example below uses ML-KEM-768 rather than an ML-DSA parameter set, because its dimensions and coefficient size give smaller numbers that are easier to follow; the method of accounting carries over unchanged to ML-DSA. The public matrix A for ML-KEM-768 has dimensions 3x3, with each polynomial having 256 coefficients of 2 bytes each, leading to a size of 3*3*256*2 = 4,608 bytes (approximately 4.5 KB) for the matrix A alone. The polynomial vectors t, s and e also contribute significantly to memory usage, with each vector requiring 3*256*2 = 1,536 bytes (approximately 1.5 KB). Hence, for a straightforward implementation, the amount of memory required is 4,608 + 3*1,536 = 9,216 bytes (approximately 9 KB). The same computation can be done for other instantiations of ML-KEM as well as for ML-DSA. ML-DSA has much higher memory requirements, both because its matrices and vectors are larger and because each coefficient occupies 4 bytes rather than 2. For ML-DSA-87, where A has dimensions 8x7, signing holds A together with the private-key vectors s1, s2 and t0 of 7, 8 and 8 elements respectively. This gives 8*7*256*4 = 57,344 bytes for A and a further 23*256*4 = 23,552 bytes for the vectors, so a straightforward implementation needs at least 79 KB of RAM during signing, before accounting for the per-iteration vectors such as y, w and z.</t>

<t>It is worth noting that different cryptographic operations may have different memory requirements. For example, during ML-DSA verification, the memory usage is lower since the private key components are not needed.</t>

<section anchor="lazy-expansion"><name>Lazy Expansion as a Memory Optimization Technique</name>

<t>The lazy expansion technique is an optimization that significantly reduces memory usage by avoiding the need to store the entire expanded matrix A in memory at once. Instead of pre-computing and storing the full matrix, lazy expansion generates parts of it on-the-fly as needed for the process. This approach leverages the fact that not all elements of the matrix are required simultaneously, allowing for a more efficient use of memory.</t>

<t>As an example, consider the computation of matrix-vector multiplication t=A*s. The matrix A is generated from a seed using an extendable-output function (XOF) <xref target="FIPS202"/>, meaning that any element of A can be computed independently when needed. Similarly, the vector s is expanded from a random seed and a nonce using a pseudo-random function (PRF).</t>

<t>Lazy expansion first initializes t with e, which can be generated directly into the t buffer. It then generates the first element of the vector s (<spanx style="verb">s(0)</spanx>), iterates over the rows of the first column of A, generates one element at a time and accumulates the products into t. It next generates <spanx style="verb">s(1)</spanx> and repeats the process for the next column, until all elements of s have been processed. For ML-KEM-768, each polynomial takes 512 bytes (256 coefficients of 2 bytes each), so only one element of s (512 bytes), one element of A (512 bytes) and the vector t (3*512 = 1,536 bytes) need to be held in memory at any time, about 2.5 KB in total compared to approximately 9 KB for a straightforward implementation. The savings are even more pronounced for ML-DSA, where combining lazy expansion with keeping the private-key vectors in packed form reduces the memory required for signing from at least 79 KB to a small fraction of that; see <xref target="Gre20"/> and <xref target="BosRS22"/> for measured figures.</t>

<t>With lazy expansion, the implementation differs slightly from the straightforward version. Also, in some cases, lazy expansion may introduce additional computational overhead. Notably, applying it to ML-DSA signing may require computing the vector y (<xref target="FIPS204"/>, Algorithm 7, line 11) twice. In this case implementers need to weigh the trade-off between memory savings and additional computation.</t>

<t>This memory optimization was initially described in <xref target="Bot19"/>.</t>

</section>
</section>
<section anchor="pre-hashing"><name>Pre-hashing as a Memory Optimization Technique</name>

<t>To address the memory consumption challenge, algorithms like ML-DSA offer a form of
pre-hash using the μ (message representative) value described in Section 6.2 of <xref target="FIPS204"/>.
The μ value provides an abstraction for pre-hashing by allowing the hash or message
representative to be computed outside the cryptographic module. This feature offers
additional flexibility by enabling the use of different cryptographic modules for the
pre-hashing step, reducing memory consumption within the cryptographic module.
The pre-computed μ value is then supplied to the cryptographic module, eliminating the need to
transmit the entire message for signing. <xref target="RFC9881"/>
discusses leveraging Externalμ-ML-DSA, where the pre-hashing step
(Externalμ-ML-DSA.Prehash) is performed in a software cryptographic module, and only the
pre-hashed message (μ) is sent to the hardware cryptographic module for signing
(Externalμ-ML-DSA.Sign). By implementing Externalμ-ML-DSA.Prehash in software and
Externalμ-ML-DSA.Sign in a hardware cryptographic module, the cryptographic workload
is efficiently distributed, making it practical for high-volume signing operations even
in memory-constrained cryptographic modules.</t>

<t>As with any pre-hashing scheme, this shifts trust to whatever component computes the digest:
the signing module never sees the original message, so it must trust that μ was faithfully
computed over the intended message. A compromised pre-hashing component could substitute μ
for a different message without the signing module detecting the substitution.</t>

<t>The main advantage of this method is that, unlike HashML-DSA, the Externalμ-ML-DSA approach
is interoperable with the standard version of ML-DSA that does not use pre-hashing. This means
a message can be signed using ML-DSA.Sign, and the verifier can independently compute μ and use
Externalμ-ML-DSA.Verify for verification - or vice versa. In both cases, the verifier
does not need to know whether the signer used internal or external pre-hashing, as the resulting
signature and verification process remain the same.</t>

</section>
</section>
<section anchor="sec-key-sizes"><name>Cryptographic Artifact Sizes for Post-Quantum Algorithms</name>

<t>The sizes of keys, ciphertexts, and signatures of post-quantum algorithms are generally larger than those of traditional
cryptographic algorithms. This increase in size is a significant consideration for
constrained devices, which often have limited memory and storage capacity. For example,
the key sizes for ML-DSA and ML-KEM are larger than those of RSA or ECDSA, which can lead to
increased memory usage and slower performance in constrained environments.</t>

<t><xref target="artifact-size"/> presents artifact sizes organized by NIST security categories published in
the initial call for proposals <xref target="NISTSecurityCategories"/>. The security categories are defined
as requiring computational resources comparable to or greater than an attack on AES (128, 192, and 256)
and SHA2/SHA3 algorithms, i.e., exhaustive key recovery for AES and optimal collision search for
SHA2/SHA3 schemes. The table lists the sizes of cryptographic artifacts for representative instantiations
of selected post-quantum cryptographic schemes of the lowest available security categories defined for
the respective schemes. X25519 and Ed25519 are included for comparison; they approximately map to NIST
Security Category 1 based on ~128-bit classical security, though this is not an official NIST designation.</t>

<texttable title="Sizes of cryptographic artifacts" anchor="artifact-size">
      <ttcol align='left'>Level</ttcol>
      <ttcol align='left'>Algorithm</ttcol>
      <ttcol align='left'>Type</ttcol>
      <ttcol align='left'>Size (bytes)</ttcol>
      <c>2</c>
      <c>ML-DSA-44</c>
      <c>Public Key</c>
      <c>1312</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Private Key</c>
      <c>2560</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Signature</c>
      <c>2420</c>
      <c>1</c>
      <c>SLH-DSA-SHA2-128s</c>
      <c>Public Key</c>
      <c>32</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Private Key</c>
      <c>64</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Signature</c>
      <c>7856</c>
      <c>1</c>
      <c>SLH-DSA-SHA2-128f</c>
      <c>Public Key</c>
      <c>32</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Private Key</c>
      <c>64</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Signature</c>
      <c>17088</c>
      <c>3</c>
      <c>LMS_SHA256_M24_H15_W4</c>
      <c>Public Key</c>
      <c>48</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Private Key</c>
      <c>44</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Signature</c>
      <c>1620</c>
      <c>3</c>
      <c>XMSS-SHA2_10_192</c>
      <c>Public Key</c>
      <c>48</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Private Key</c>
      <c>104</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Signature</c>
      <c>1492</c>
      <c>1</c>
      <c>ML-KEM-512</c>
      <c>Public Key</c>
      <c>800</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Private Key</c>
      <c>1632</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Ciphertext</c>
      <c>768</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Shared Secret</c>
      <c>32</c>
      <c>1*</c>
      <c>X25519</c>
      <c>Public Key</c>
      <c>32</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Private Key</c>
      <c>32</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Shared Secret</c>
      <c>32</c>
      <c>1*</c>
      <c>Ed25519</c>
      <c>Public Key</c>
      <c>32</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Private Key</c>
      <c>32</c>
      <c>&#160;</c>
      <c>&#160;</c>
      <c>Signature</c>
      <c>64</c>
</texttable>

<t>Corresponding sizes for higher security categories will typically be larger - see <xref target="FIPS203"/>, <xref target="FIPS204"/>, <xref target="FIPS205"/>, <xref target="SP800-208"/>, <xref target="RFC9858"/> for sizes for all parameter sets.</t>

</section>
<section anchor="sig-perf"><name>Optimizing Performance in PQC Signature Schemes</name>

<t>When implementing PQC signature algorithms in constrained cryptographic modules,
performance optimization becomes a critical consideration. Transmitting the entire message
to the cryptographic module for signing can lead to significant overhead, especially for
large payloads. To address this, implementers can leverage techniques that reduce the data
transmitted to the cryptographic module, thereby improving efficiency and scalability.</t>

<t>One effective approach involves sending only a message digest to the cryptographic module
for signing. By signing the digest of the content rather than the entire content, the
communication between the application and the cryptographic module is minimized, enabling
better performance. This method is applicable for both ML-DSA and SLH-DSA. For such algorithms,
a mechanism is often provided to pre-hash or process the message in a way that avoids sending
the entire raw message for signing. In particular, algorithms like SLH-DSA present challenges due to
their construction, which requires two passes over the message during the
signing process. The signer must therefore either retain the message for the second pass
or receive it twice. This differs from traditional algorithms like RSA or ECDSA,
which allow for more efficient processing of the message, without requiring multiple
passes or intermediate processing of the digest.</t>

<section anchor="mldsa-rej-sampling"><name>Impact of rejection sampling in ML-DSA Signing on performance</name>

<t>In constrained and battery-powered IoT devices that perform ML-DSA signing, the rejection-sampling
loop introduces variability in signing latency and energy consumption due to the probabilistic
nature of the signing process. While this results in a variable number of iterations in the signing
algorithm, the expected number of attempts for the standardized ML-DSA parameter sets is quantified
below.</t>

<t>The analysis in this section follows the algorithmic structure and assumptions defined in
<xref target="FIPS204"/>. The results characterize the expected behavior of ML-DSA rather than any particular
implementation.</t>

<t>The ML-DSA signature scheme uses the Fiat-Shamir with Aborts construction <xref target="Lyu09"/>. As a
result, the signature generation algorithm is built around a rejection-sampling loop. This
section examines the rejection-sampling behavior of ML-DSA, as rejection sampling is not
commonly used as a core mechanism in traditional digital signature schemes.</t>

<t>Rejection sampling is used to ensure that intermediate and output values follow the
distributions required by the security proof. In particular, after computing candidate signature
components, the signer checks whether certain norm bounds are satisfied. If any of these bounds
are violated, the entire signing attempt is discarded and restarted with fresh pseudorandom values.</t>

<t>The purpose of rejection sampling is twofold: First, it prevents leakage of information about the
secret key through out-of-range values that could otherwise bias the distribution of signatures.
Second, it ensures that the distribution of valid signatures is statistically close to the ideal
distribution assumed in the security reduction, namely the zero-knowledge property underlying the
reduction to the SelfTargetMSIS problem (see Section 6.2.1 of <xref target="Dilithium"/>).</t>

<t>The number of rejections during signature generation depends on three factors:</t>

<t><list style="symbols">
  <t>the message representative μ, which depends on the message, the context string (see <xref target="FIPS204"/>, Section 5.2) and the public key</t>
  <t>the secret key material</t>
  <t>when hedged signing is used (see <xref target="FIPS204"/>, Section 3.4), the random seed</t>
</list></t>

<t>As a result, some message-key combinations may lead to a higher number of
rejection iterations than others.</t>

<t>Each signing attempt can be modeled as an independent Bernoulli trial: an attempt
either succeeds or is rejected, with a fixed per-attempt acceptance probability p.
Under this assumption, the number of attempts until success follows a geometric
distribution, and the expected number of attempts is 1/p.</t>

<t>The values below are taken from <xref target="KWI2026"/>, assuming a random bit generator
(RBG) as specified in <xref target="FIPS204"/> (Section 3.6.1).</t>

<texttable title="Per-attempt acceptance probability and expected number of attempts for the given ML-DSA variant." anchor="Expected_Attempts">
      <ttcol align='left'>ML-DSA Variant</ttcol>
      <ttcol align='left'>Per-attempt Acceptance</ttcol>
      <ttcol align='left'>Expected Number of Attempts</ttcol>
      <c>ML-DSA-44</c>
      <c>0.2293</c>
      <c>4.361</c>
      <c>ML-DSA-65</c>
      <c>0.1947</c>
      <c>5.137</c>
      <c>ML-DSA-87</c>
      <c>0.2561</c>
      <c>3.905</c>
</texttable>

<t>The cumulative distribution function (CDF) follows directly from the geometric
model. <xref target="MLDSA_Sign_CDF"/> shows, for each ML-DSA variant, the probability that
signing completes within a given number of iterations. The first rows matter
most in practice: more than half of all signing operations complete within 3
iterations (4 for ML-DSA-65).</t>

<texttable title="Probability of completing the signing process within the given number of iterations, for each ML-DSA variant." anchor="MLDSA_Sign_CDF">
      <ttcol align='left'>Iteration</ttcol>
      <ttcol align='left'>ML-DSA-44</ttcol>
      <ttcol align='left'>ML-DSA-65</ttcol>
      <ttcol align='left'>ML-DSA-87</ttcol>
      <c>1</c>
      <c>0.2293</c>
      <c>0.1947</c>
      <c>0.2561</c>
      <c>2</c>
      <c>0.4060</c>
      <c>0.3514</c>
      <c>0.4466</c>
      <c>3</c>
      <c>0.5423</c>
      <c>0.4777</c>
      <c>0.5883</c>
      <c>4</c>
      <c>0.6472</c>
      <c>0.5794</c>
      <c>0.6937</c>
      <c>5</c>
      <c>0.7281</c>
      <c>0.6612</c>
      <c>0.7722</c>
      <c>6</c>
      <c>0.7905</c>
      <c>0.7272</c>
      <c>0.8305</c>
      <c>7</c>
      <c>0.8385</c>
      <c>0.7803</c>
      <c>0.8739</c>
      <c>8</c>
      <c>0.8755</c>
      <c>0.8231</c>
      <c>0.9062</c>
      <c>9</c>
      <c>0.9041</c>
      <c>0.8575</c>
      <c>0.9302</c>
      <c>10</c>
      <c>0.9261</c>
      <c>0.8852</c>
      <c>0.9481</c>
      <c>11</c>
      <c>0.9430</c>
      <c>0.9076</c>
      <c>0.9614</c>
      <c>12</c>
      <c>0.9561</c>
      <c>0.9256</c>
      <c>0.9713</c>
</texttable>

<t>Inverting the CDF gives the minimum number of iterations n required to reach a
desired completion probability, n &gt;= ln(1 - target) / ln(1 - p). This is the
figure implementations need when budgeting for worst-case latency or energy
rather than for the average case.</t>

<texttable title="Iterations required to reach a given probability of completing the signing process, for each ML-DSA variant." anchor="MLDSA_Sign_Quantiles">
      <ttcol align='left'>Target</ttcol>
      <ttcol align='left'>ML-DSA-44</ttcol>
      <ttcol align='left'>ML-DSA-65</ttcol>
      <ttcol align='left'>ML-DSA-87</ttcol>
      <c>90%</c>
      <c>9</c>
      <c>11</c>
      <c>8</c>
      <c>95%</c>
      <c>12</c>
      <c>14</c>
      <c>11</c>
      <c>99%</c>
      <c>18</c>
      <c>22</c>
      <c>16</c>
</texttable>

<t>Every variant reaches at least 90% within 11 iterations, but the tail is long:
ML-DSA-65 needs 22 iterations for 99%, against an expected 5.1.</t>

<t><xref target="FIPS204"/> Appendix C bounds the signing loop at 814 iterations for a
failure probability of at most 2^-256, based on the expected repetition
counts in <xref target="FIPS204"/> Table 1. A more precise computation of these counts
(see <xref target="Expected_Attempts"/>) gives 5.137 for ML-DSA-65, the parameter set
with the highest repetition count, which yields a limit of 820 iterations;
with 814, the probability that signing fails to complete is about
2^-254.2, slightly short of the 2^-256 target. This does not affect the
security of ML-DSA, as such a failure only requires signing to be
retried. The FIPS 204 potential updates <xref target="FIPS204_errata"/> also conclude
that 814 is too low, but compute the limit from the rounded count 5.14,
giving 821. This is one iteration above the minimum derived here, so it
also meets the 2^-256 target. For FIPS compliance, implementations that
bound the loop should use the limit specified in <xref target="FIPS204"/>, or in a
published update to it.</t>

<section anchor="practical-implications-for-constrained-cryptographic-modules"><name>Practical Implications for Constrained Cryptographic Modules</name>

<t>As shown above, the rejection-sampling loop in ML-DSA signing leads to a probabilistic runtime
with a geometrically distributed number of iterations. While the expected execution time is
small, the tail of the distribution implies that, with low probability, a signing operation
may require significantly more iterations than average. This unfavorable tail behavior represents
a practical concern for ML-DSA deployments on constrained devices with limited execution
capability and may require additional consideration.</t>

<t>As discussed in <xref target="Seed"/>, in many deployment scenarios, constrained devices primarily perform signature verification, while signature generation is performed on more capable systems (e.g., firmware signing infrastructure). Therefore, the impact of rejection sampling is primarily relevant for devices that perform ML-DSA signing.</t>

<t>Devices that only verify signatures are not affected, as those operations do not involve rejection sampling and have deterministic execution times.</t>

</section>
<section anchor="suggestions-for-benchmarking-ml-dsa-signing-performance"><name>Suggestions for benchmarking ML-DSA Signing Performance</name>

<t>When benchmarking ML-DSA signing performance in constrained cryptographic modules, it is
important to account for the probabilistic nature of the rejection-sampling loop. Reporting
only a single timing measurement or a best-case execution time may lead to misleading conclusions
about practical performance.</t>

<t>Libraries implementing ML-DSA should provide a mechanism to report the number of
rejection-sampling iterations used during the most recent signing operation. This enables
benchmarking tools to accurately compute average signing times across multiple signing operations.</t>

<t>To provide a more comprehensive assessment of ML-DSA signing performance, benchmarks may report
all or some of the following metrics:</t>

<t><list style="numbers" type="1">
  <t>Single-iteration signing time:
The signing time for a signature operation that completes within a single iteration of the
rejection-sampling loop. This metric includes the fixed setup cost incurred once per signing
call (such as matrix expansion, message digest computation, and similar precomputations), plus
the cost of one loop operation. It reflects the best-case performance of the signing algorithm
and provides insight into the efficiency of the core signing operation.</t>
  <t>Average signing time:
Since the iteration count follows a geometric distribution (as described in <xref target="mldsa-rej-sampling"/>),
the expected signing time can be computed analytically as the fixed setup cost plus the per-iteration
cost multiplied by the expected number of iterations from <xref target="Expected_Attempts"/>.
Implementations may also measure average signing time empirically over a sufficiently large number of
signing operations, using independent messages and, where applicable, independent randomness, to validate
against the analytical model on the target hardware.</t>
</list></t>

<t>Rather than relying on ad hoc random inputs, benchmarks may use a standardized input data set covering
best-case, average, and worst-case vectors with documented occurrence probabilities, to ensure
reproducibility and comparability across implementations. This requires identifying message, key and randomness
combinations that result in the target iteration counts, e.g., a single iteration for the best case, the
expected count rounded to the nearest integer for the average case, and a high quantile such as the 99% value
from <xref target="MLDSA_Sign_Quantiles"/> for the worst case.</t>

</section>
</section>
</section>
<section anchor="additional-considerations"><name>Additional Considerations for PQC Use in Constrained Devices</name>

<section anchor="key-rotation-and-renewal"><name>Key Rotation and Renewal</name>

<t>In constrained devices, managing the lifecycle of cryptographic
keys including periodic key rotation and renewal is critical for maintaining long-term
security and supporting cryptographic agility. While constrained devices may rely on
dedicated key-storage hardware for secure key storage and operations, the
responsibility for orchestrating key rotation typically resides in the application layer
or external device management infrastructure.</t>

<t>Although the underlying cryptographic module may offer primitives to securely generate new
key pairs, store fresh seeds, or delete obsolete keys, these capabilities must be
integrated into the device's broader key management framework. This process is especially
important in the context of PQC, where evolving research may lead to changes in
recommended algorithms, parameters, and key management practices.</t>

<t>The security of PQC schemes continues to evolve, with potential risks arising from
advances in post-quantum algorithms, cryptanalytic or implementation vulnerabilities. As a
result, constrained devices should be designed to support flexible and updatable key
management policies. This includes the ability to:</t>

<t><list style="symbols">
  <t>Rotate keys periodically to limit the impact of a key compromise,</t>
  <t>Update algorithm choices or key sizes based on emerging security guidance,</t>
  <t>Reconfigure cryptographic profile of the device via firmware updates.</t>
</list></t>

</section>
</section>
<section anchor="firmware"><name>Post-quantum Firmware Upgrades for Constrained Devices</name>

<t>Constrained devices deployed in the field require periodic firmware upgrades to patch
security vulnerabilities, introduce new cryptographic algorithms, and improve overall
functionality. However, if not designed to withstand attacks from a CRQC, the firmware update
process itself can become a critical attack vector. If an adversary compromises the update mechanism,
they could introduce malicious firmware, undermining all other security properties of the
cryptographic modules. Therefore, ensuring a post-quantum firmware upgrade process is critical
for the security of deployed constrained devices.</t>

<t>CRQCs pose an additional risk by breaking traditional digital signatures (e.g., RSA,
ECDSA) used to authenticate firmware updates. If firmware verification relies on
traditional signature algorithms, attackers could generate forged signatures in the future
and distribute malicious updates.</t>

<section anchor="post-quantum-firmware-authentication"><name>Post-Quantum Firmware Authentication</name>

<t>To ensure the integrity and authenticity of firmware updates, constrained devices will have to adopt PQC digital signature schemes for code signing.
These algorithms must provide long-term security, operate efficiently in low-resource environments, and be compatible with secure update mechanisms, such as the firmware update architecture for IoT described in <xref target="RFC9019"/>.</t>

<t><xref target="I-D.ietf-suit-mti"/> defines mandatory-to-implement cryptographic algorithms for IoT devices, and recommends use of HSS/LMS <xref target="RFC8554"/> to secure software updates. The IETF working groups may consider adding post-quantum algorithms, such as SLH-DSA and ML-DSA, in future specifications.</t>

<t>Stateful hash-based signature schemes, such as HSS/LMS or the similar XMSS <xref target="RFC8391"/>, are good candidates for signing firmware updates. Those schemes offer efficient verification times, making them more practical choices for constrained environments where performance and memory usage are key concerns.
Their security is based on the security of the underlying hash function, which is well-understood.
A major downside of stateful hash-based signatures is the requirement to keep track of which One-Time Signature (OTS) keys have been used, since reuse of a single OTS key allows for signature forgeries.
However, in the case of firmware updates, the OTS keys will be signing versioned updates, which may make state management easier.
<xref target="RFC10033"/> discusses various strategies for a correct state and backup management for stateful hash-based signatures.</t>

<t>Other post-quantum signature algorithms may also be viable for firmware signing:</t>

<t><list style="symbols">
  <t>SLH-DSA, a stateless hash-based signature specified in <xref target="FIPS205"/>, also has well-understood security based on the security of its underlying hash function, and additionally doesn't have the complexities associated with state management that HSS and XMSS have.</t>
</list></t>

<t>However, signature generation and verification are comparatively slow, and signature sizes are generally larger than other post-quantum algorithms.
SLH-DSA's suitability as a firmware signing algorithm will depend on the capabilities of the underlying hardware.</t>

<t><list style="symbols">
  <t>ML-DSA is a lattice-based signature algorithm specified in <xref target="FIPS204"/>.
It is more performant than SLH-DSA, with significantly faster signing and verification times, as well as shorter signatures.</t>
</list></t>

<t>This will make it possible to implement on a wider range of constrained devices.
The mathematical problem underpinning ML-DSA, Module Learning With Errors (M-LWE), is believed to be a hard problem by the cryptographic community, and hence ML-DSA is believed to be secure.
Cryptographers are more confident still in the security of hash-based signatures than M-LWE, so developers may wish to factor that in when choosing a firmware signing algorithm.</t>

<t>In firmware update and secure boot scenarios, signature verification is typically performed during early boot stages, where the bootloader has exclusive access to system resources. In such environments, the practical impact of resource constraints on signature verification is reduced compared to general runtime environments.</t>

<t>Verification does not always occur during early boot. In devices that keep a second firmware image and switch to it only after verifying it, the new image is verified while the current firmware runs, so verification competes with the device's normal workload for CPU, RAM, and energy. The optimizations in <xref target="sig-mem"/> and <xref target="sig-perf"/> are therefore especially relevant when verification runs concurrently with normal operation.</t>

</section>
<section anchor="hybrid-signature-approaches"><name>Hybrid Signature Approaches</name>

<t>To enable secure migration from traditional to post-quantum security, PQ/T hybrid digital signature methods can be used for firmware authentication, combining a traditional and a post-quantum algorithm using either non-composite or composite constructions as defined in <xref target="RFC9794"/>.</t>

<t>A non-composite approach, where both signatures are generated and carried separately, is simple to implement, requires minimal changes to existing signing, and aligns well with current secure boot and update architectures.</t>

<t>Composite constructions, which combine multiple algorithms into a single signature, require changes to cryptographic processing. In such constructions, the additional cost of including a traditional algorithm is typically small compared to the post-quantum component, and overall resource usage remains dominated by the post-quantum algorithm, particularly in terms of key size, signature size, code size, and verification cost.</t>

<t>Implementations should ensure that verification enforces the intended hybrid authentication property, namely that authentication remains secure as long as at least one component algorithm remains secure.</t>

</section>
</section>
<section anchor="iana-considerations"><name>IANA Considerations</name>

<t>This document requires no IANA actions.</t>

</section>
<section anchor="security-considerations"><name>Security Considerations</name>

<t>The security considerations for key management in constrained devices for PQC focus on the
secure storage and handling of cryptographic seeds, which are used to derive private keys.
Seeds must be protected with the same security measures as private keys, and key
derivation should be efficient and secure within resource-constrained cryptographic
module. Secure export and backup mechanisms for seeds are essential to ensure recovery in
case of hardware failure, but the exported seeds must be encrypted and protected from
unauthorized access.</t>

<t>Key generation, encapsulation, and hedged signing require random values suitable for cryptographic use. An inadequate random bit generator (RBG) can allow an attacker to recover a private key by searching a small set of possibilities rather than the whole key space. Implementations should meet the RBG requirements in <xref target="FIPS203"/>, <xref target="FIPS204"/>, <xref target="FIPS205"/>, and <xref target="SP800-208"/>.</t>

<section anchor="side-channel-protection"><name>Side Channel Protection</name>

<t>Side-channel attacks exploit physical leaks during cryptographic operations, such as timing information, power consumption, electromagnetic emissions, or other physical characteristics, to extract sensitive data like private keys or seeds. Given the sensitivity of the seed and private key in PQC key generation, it is critical to consider side-channel protection in cryptographic module design. While side-channel attacks remain an active research topic, they are a major concern in secure hardware design and must not be overlooked. Cryptographic modules must incorporate strong countermeasures against side-channel vulnerabilities to prevent attackers from gaining insights into secret data during cryptographic operations.</t>

<t>ML-DSA and SLH-DSA support both deterministic and hedged signing. On platforms where side-channel attacks are a concern and cannot be otherwise mitigated, hedged signing should be used, as discussed in Section 3.4 of <xref target="FIPS204"/> and Section 9.2 of <xref target="FIPS205"/>.</t>

</section>
</section>
<section anchor="acknowledgments"><name>Acknowledgments</name>

<t>Thanks to Jean-Pierre Fiset, Richard Kettlewell, Mike Ounsworth, Russ Housley, Keegan Dasilva Barbosa, Hannes Tschofenig, Aritra Banerjee, Mohamed Boucadair, Ketan Talaulikar and Ionut Mihalcea for the detailed review.</t>

</section>


  </middle>

  <back>


<references title='References' anchor="sec-combined-references">

    <references title='Normative References' anchor="sec-normative-references">



<reference anchor="FIPS203">
  <front>
    <title>Module-lattice-based key-encapsulation mechanism standard</title>
    <author>
      <organization/>
    </author>
    <date month="August" year="2024"/>
  </front>
  <seriesInfo name="DOI" value="10.6028/nist.fips.203"/>
<refcontent>National Institute of Standards and Technology (U.S.)</refcontent></reference>
<reference anchor="FIPS204">
  <front>
    <title>Module-lattice-based digital signature standard</title>
    <author>
      <organization/>
    </author>
    <date month="August" year="2024"/>
  </front>
  <seriesInfo name="DOI" value="10.6028/nist.fips.204"/>
<refcontent>National Institute of Standards and Technology (U.S.)</refcontent></reference>
<reference anchor="FIPS205">
  <front>
    <title>Stateless hash-based digital signature standard</title>
    <author>
      <organization/>
    </author>
    <date month="August" year="2024"/>
  </front>
  <seriesInfo name="DOI" value="10.6028/nist.fips.205"/>
<refcontent>National Institute of Standards and Technology (U.S.)</refcontent></reference>

<reference anchor="I-D.ietf-iotops-7228bis">
   <front>
      <title>Terminology for Constrained-Node Networks</title>
      <author fullname="Carsten Bormann" initials="C." surname="Bormann">
         <organization>Universität Bremen TZI</organization>
      </author>
      <author fullname="Mehmet Ersue" initials="M." surname="Ersue">
         </author>
      <author fullname="Ari Keränen" initials="A." surname="Keränen">
         <organization>Ericsson</organization>
      </author>
      <author fullname="Carles Gomez" initials="C." surname="Gomez">
         <organization>Universitat Politecnica de Catalunya</organization>
      </author>
      <date day="6" month="July" year="2026"/>
      <abstract>
	 <t>   The Internet Protocol Suite is increasingly used on small devices
   with severe constraints on power, memory, and processing resources,
   creating constrained-node networks.  This document provides a number
   of basic terms that have been useful in research and standardization
   work for constrained-node networks.

   This document obsoletes RFC 7228.

	 </t>
      </abstract>
   </front>
   <seriesInfo name="Internet-Draft" value="draft-ietf-iotops-7228bis-10"/>

</reference>
<reference anchor="RFC8554">
  <front>
    <title>Leighton-Micali Hash-Based Signatures</title>
    <author fullname="D. McGrew" initials="D." surname="McGrew"/>
    <author fullname="M. Curcio" initials="M." surname="Curcio"/>
    <author fullname="S. Fluhrer" initials="S." surname="Fluhrer"/>
    <date month="April" year="2019"/>
    <abstract>
      <t>This note describes a digital-signature system based on cryptographic hash functions, following the seminal work in this area of Lamport, Diffie, Winternitz, and Merkle, as adapted by Leighton and Micali in 1995. It specifies a one-time signature scheme and a general signature scheme. These systems provide asymmetric authentication without using large integer mathematics and can achieve a high security level. They are suitable for compact implementations, are relatively simple to implement, and are naturally resistant to side-channel attacks. Unlike many other signature systems, hash-based signatures would still be secure even if it proves feasible for an attacker to build a quantum computer.</t>
      <t>This document is a product of the Crypto Forum Research Group (CFRG) in the IRTF. This has been reviewed by many researchers, both in the research group and outside of it. The Acknowledgements section lists many of them.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="8554"/>
  <seriesInfo name="DOI" value="10.17487/RFC8554"/>
</reference>
<reference anchor="RFC8391">
  <front>
    <title>XMSS: eXtended Merkle Signature Scheme</title>
    <author fullname="A. Huelsing" initials="A." surname="Huelsing"/>
    <author fullname="D. Butin" initials="D." surname="Butin"/>
    <author fullname="S. Gazdag" initials="S." surname="Gazdag"/>
    <author fullname="J. Rijneveld" initials="J." surname="Rijneveld"/>
    <author fullname="A. Mohaisen" initials="A." surname="Mohaisen"/>
    <date month="May" year="2018"/>
    <abstract>
      <t>This note describes the eXtended Merkle Signature Scheme (XMSS), a hash-based digital signature system that is based on existing descriptions in scientific literature. This note specifies Winternitz One-Time Signature Plus (WOTS+), a one-time signature scheme; XMSS, a single-tree scheme; and XMSS^MT, a multi-tree variant of XMSS. Both XMSS and XMSS^MT use WOTS+ as a main building block. XMSS provides cryptographic digital signatures without relying on the conjectured hardness of mathematical problems. Instead, it is proven that it only relies on the properties of cryptographic hash functions. XMSS provides strong security guarantees and is even secure when the collision resistance of the underlying hash function is broken. It is suitable for compact implementations, is relatively simple to implement, and naturally resists side-channel attacks. Unlike most other signature systems, hash-based signatures can so far withstand known attacks using quantum computers.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="8391"/>
  <seriesInfo name="DOI" value="10.17487/RFC8391"/>
</reference>



    </references>

    <references title='Informative References' anchor="sec-informative-references">



<reference anchor="SP800-208">
  <front>
    <title>Recommendation for Stateful Hash-Based Signature Schemes</title>
    <author fullname="David A. Cooper" initials="D." surname="Cooper">
      <organization/>
    </author>
    <author fullname="Daniel C. Apon" initials="D." surname="Apon">
      <organization/>
    </author>
    <author fullname="Quynh H. Dang" initials="Q." surname="Dang">
      <organization/>
    </author>
    <author fullname="Michael S. Davidson" initials="M." surname="Davidson">
      <organization/>
    </author>
    <author fullname="Morris J. Dworkin" initials="M." surname="Dworkin">
      <organization/>
    </author>
    <author fullname="Carl A. Miller" initials="C." surname="Miller">
      <organization/>
    </author>
    <date month="October" year="2020"/>
  </front>
  <seriesInfo name="DOI" value="10.6028/nist.sp.800-208"/>
<refcontent>National Institute of Standards and Technology</refcontent></reference>
<reference anchor="FIPS202">
  <front>
    <title>SHA-3 standard :: permutation-based hash and extendable-output functions</title>
    <author>
      <organization/>
    </author>
    <date year="2015"/>
  </front>
  <seriesInfo name="DOI" value="10.6028/nist.fips.202"/>
<refcontent>National Institute of Standards and Technology (U.S.)</refcontent></reference>

<reference anchor="ISO19790" target="https://www.iso.org/standard/82423.html">
  <front>
    <title>Information security, cybersecurity, and privacy protection - Security requirements for cryptographic modules</title>
    <author >
      <organization>ISO</organization>
    </author>
    <date year="2025" month="February"/>
  </front>
</reference>
<reference anchor="BIND" target="https://eprint.iacr.org/2024/523.pdf">
  <front>
    <title>Unbindable Kemmy Schmidt: ML-KEM is neither MAL-BIND-K-CT nor MAL-BIND-K-PK</title>
    <author initials="S." surname="Schmieg">
      <organization></organization>
    </author>
    <date year="2024" month="April"/>
  </front>
</reference>
<reference anchor="HQC" target="https://pqc-hqc.org/doc/hqc_specifications_2025_08_22.pdf">
  <front>
    <title>Hamming Quasi-Cyclic (HQC)</title>
    <author initials="" surname="Gaborit et al">
      <organization></organization>
    </author>
    <date year="2025" month="August"/>
  </front>
</reference>
<reference anchor="Falcon" target="https://falcon-sign.info/falcon.pdf">
  <front>
    <title>Falcon: Fast-Fourier Lattice-based Compact Signatures over NTRU</title>
    <author initials="P.-A." surname="Fouque">
      <organization></organization>
    </author>
    <author initials="J." surname="Hoffstein">
      <organization></organization>
    </author>
    <author initials="P." surname="Kirchner">
      <organization></organization>
    </author>
    <author initials="V." surname="Lyubashevsky">
      <organization></organization>
    </author>
    <author initials="T." surname="Pornin">
      <organization></organization>
    </author>
    <author initials="T." surname="Prest">
      <organization></organization>
    </author>
    <author initials="T." surname="Ricosset">
      <organization></organization>
    </author>
    <author initials="G." surname="Seiler">
      <organization></organization>
    </author>
    <author initials="W." surname="Whyte">
      <organization></organization>
    </author>
    <author initials="Z." surname="Zhang">
      <organization></organization>
    </author>
    <date year="2020" month="October"/>
  </front>
</reference>
<reference anchor="Stream-SPHINCS" target="https://eprint.iacr.org/2021/1072.pdf">
  <front>
    <title>Streaming SPHINCS+ for Embedded Devices using the Example of TPMs</title>
    <author initials="R." surname="Niederhagen">
      <organization></organization>
    </author>
    <author initials="J." surname="Roth">
      <organization></organization>
    </author>
    <author initials="J." surname="Walde">
      <organization></organization>
    </author>
    <date year="2021" month="August"/>
  </front>
</reference>
<reference anchor="BosRS22" target="https://eprint.iacr.org/2022/323.pdf">
  <front>
    <title>Dilithium for Memory Constrained Devices</title>
    <author initials="J." surname="Bos">
      <organization></organization>
    </author>
    <author initials="J." surname="Renes">
      <organization></organization>
    </author>
    <author initials="A." surname="Sprenkels">
      <organization></organization>
    </author>
    <date year="2022" month="December"/>
  </front>
</reference>


<reference anchor="REC-KEM">
  <front>
    <title>Recommendations for key-encapsulation mechanisms</title>
    <author fullname="Gorjan Alagic" initials="G." surname="Alagic">
      <organization/>
    </author>
    <author fullname="Elaine Barker" initials="E." surname="Barker">
      <organization/>
    </author>
    <author fullname="Lily Chen" initials="L." surname="Chen">
      <organization/>
    </author>
    <author fullname="Dustin Moody" initials="D." surname="Moody">
      <organization/>
    </author>
    <author fullname="Angela Robinson" initials="A." surname="Robinson">
      <organization/>
    </author>
    <author fullname="Hamilton Silberg" initials="H." surname="Silberg">
      <organization/>
    </author>
    <author fullname="Noah Waller" initials="N." surname="Waller">
      <organization/>
    </author>
    <date month="September" year="2025"/>
  </front>
  <seriesInfo name="DOI" value="10.6028/nist.sp.800-227"/>
<refcontent>National Institute of Standards and Technology (U.S.)</refcontent></reference>
<reference anchor="Lyu09">
  <front>
    <title>Fiat-Shamir with Aborts: Applications to Lattice and Factoring-Based Signatures</title>
    <author fullname="Vadim Lyubashevsky" initials="V." surname="Lyubashevsky">
      <organization/>
    </author>
    <date year="2009"/>
  </front>
  <seriesInfo name="Lecture Notes in Computer Science" value="pp. 598-616"/>
  <seriesInfo name="DOI" value="10.1007/978-3-642-10366-7_35"/>
  <seriesInfo name="ISBN" value="[&quot;9783642103650&quot;, &quot;9783642103667&quot;]"/>
<refcontent>Springer Berlin Heidelberg</refcontent></reference>

<reference anchor="Dilithium" target="https://pq-crystals.org/dilithium/data/dilithium-specification-round3-20210208.pdf">
  <front>
    <title>CRYSTALS-Dilithium: Algorithm Specifications and Supporting Documentation (Version 3.1)</title>
    <author initials="S." surname="Bai">
      <organization></organization>
    </author>
    <author initials="L." surname="Ducas">
      <organization></organization>
    </author>
    <author initials="E." surname="Kiltz">
      <organization></organization>
    </author>
    <author initials="T." surname="Lepoint">
      <organization></organization>
    </author>
    <author initials="V." surname="Lyubashevsky">
      <organization></organization>
    </author>
    <author initials="P." surname="Schwabe">
      <organization></organization>
    </author>
    <author initials="G." surname="Seiler">
      <organization></organization>
    </author>
    <author initials="D." surname="Stehle">
      <organization></organization>
    </author>
    <date year="2021" month="February"/>
  </front>
</reference>
<reference anchor="PKCS11" target="https://docs.oasis-open.org/pkcs11/pkcs11-spec/v3.2/os/">
  <front>
    <title>PKCS #11 Specification Version 3.2</title>
    <author >
      <organization>OASIS</organization>
    </author>
    <date year="2026" month="June"/>
  </front>
</reference>
<reference anchor="NISTSecurityCategories" target="https://csrc.nist.gov/projects/post-quantum-cryptography/post-quantum-cryptography-standardization/evaluation-criteria/security-(evaluation-criteria)">
  <front>
    <title>Post-Quantum Cryptography: Security (Evaluation Criteria)</title>
    <author >
      <organization>NIST</organization>
    </author>
    <date year="2017" month="January"/>
  </front>
</reference>
<reference anchor="Bot19" target="https://eprint.iacr.org/2019/489.pdf">
  <front>
    <title>Memory-Efficient High-Speed Implementation of Kyber on Cortex-M4</title>
    <author initials="L." surname="Botros">
      <organization></organization>
    </author>
    <author initials="M. J." surname="Kannwischer">
      <organization></organization>
    </author>
    <author initials="P." surname="Schwabe">
      <organization></organization>
    </author>
    <date year="2019" month="May"/>
  </front>
</reference>


<reference anchor="Gre20">
  <front>
    <title>Compact Dilithium Implementations on Cortex-M3 and Cortex-M4</title>
    <author fullname="Denisa O. C. Greconici" initials="D." surname="Greconici">
      <organization/>
    </author>
    <author fullname="Matthias J. Kannwischer" initials="M." surname="Kannwischer">
      <organization/>
    </author>
    <author fullname="Daan Sprenkels" initials="D." surname="Sprenkels">
      <organization/>
    </author>
    <date month="December" year="2020"/>
  </front>
  <seriesInfo name="IACR Transactions on Cryptographic Hardware and Embedded Systems" value="pp. 1-24"/>
  <seriesInfo name="DOI" value="10.46586/tches.v2021.i1.1-24"/>
<refcontent>Universitatsbibliothek der Ruhr-Universitat Bochum</refcontent></reference>

<reference anchor="Smaller-SPHINCS" target="https://eprint.iacr.org/2024/018.pdf">
  <front>
    <title>Smaller Sphincs+ or, Honey, I Shrunk the Signatures</title>
    <author initials="S." surname="Fluhrer">
      <organization></organization>
    </author>
    <author initials="Q." surname="Dang">
      <organization></organization>
    </author>
    <date year="2024" month="January"/>
  </front>
</reference>
<reference anchor="IEEE802.1AR" target="https://standards.ieee.org/ieee/802.1AR/6995/">
  <front>
    <title>IEEE Standard for Local and Metropolitan Area Networks - Secure Device Identity</title>
    <author >
      <organization>IEEE</organization>
    </author>
    <date year="2018"/>
  </front>
  <seriesInfo name="IEEE" value="802.1AR-2018"/>
</reference>
<reference anchor="KWI2026" target="https://eprint.iacr.org/2026/2456">
  <front>
    <title>On the Expected Number of Iterations in ML-DSA Signing: A Complete Accounting of the Rejection Conditions</title>
    <author initials="K." surname="Kwiatkowski">
      <organization></organization>
    </author>
    <date year="2026" month="October"/>
  </front>
</reference>
<reference anchor="FIPS204_errata" target="https://csrc.nist.gov/files/pubs/fips/204/final/docs/fips-204-potential-updates.xlsx">
  <front>
    <title>FIPS 204 - Potential Updates (Errata)</title>
    <author >
      <organization>NIST</organization>
    </author>
    <date year="2026" month="July"/>
  </front>
</reference>
<reference anchor="KWI2026CODE" target="https://doi.org/10.5281/zenodo.24005330">
  <front>
    <title>On the Expected Number of Iterations in ML-DSA Signing: A Complete Accounting of the Rejection Conditions (Code)</title>
    <author initials="K." surname="Kwiatkowski">
      <organization></organization>
    </author>
    <date year="2026" month="October"/>
  </front>
</reference>


<reference anchor="RFC9881">
  <front>
    <title>Internet X.509 Public Key Infrastructure -- Algorithm Identifiers for the Module-Lattice-Based Digital Signature Algorithm (ML-DSA)</title>
    <author fullname="J. Massimo" initials="J." surname="Massimo"/>
    <author fullname="P. Kampanakis" initials="P." surname="Kampanakis"/>
    <author fullname="S. Turner" initials="S." surname="Turner"/>
    <author fullname="B. E. Westerbaan" initials="B. E." surname="Westerbaan"/>
    <date month="October" year="2025"/>
    <abstract>
      <t>Digital signatures are used within X.509 certificates and Certificate Revocation Lists (CRLs), and to sign messages. This document specifies the conventions for using FIPS 204, the Module-Lattice-Based Digital Signature Algorithm (ML-DSA) in Internet X.509 certificates and CRLs. The conventions for the associated signatures, subject public keys, and private key are also described.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9881"/>
  <seriesInfo name="DOI" value="10.17487/RFC9881"/>
</reference>
<reference anchor="RFC9935">
  <front>
    <title>Internet X.509 Public Key Infrastructure - Algorithm Identifiers for the Module-Lattice-Based Key-Encapsulation Mechanism (ML-KEM)</title>
    <author fullname="S. Turner" initials="S." surname="Turner"/>
    <author fullname="P. Kampanakis" initials="P." surname="Kampanakis"/>
    <author fullname="J. Massimo" initials="J." surname="Massimo"/>
    <author fullname="B. E. Westerbaan" initials="B. E." surname="Westerbaan"/>
    <date month="March" year="2026"/>
    <abstract>
      <t>The Module-Lattice-Based Key-Encapsulation Mechanism (ML-KEM) is a quantum-resistant Key Encapsulation Mechanism. This document specifies the conventions for using the ML-KEM in X.509 Public Key Infrastructure. The conventions for the subject public keys and private keys are also specified.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9935"/>
  <seriesInfo name="DOI" value="10.17487/RFC9935"/>
</reference>
<reference anchor="RFC9958">
  <front>
    <title>Post-Quantum Cryptography for Engineers</title>
    <author fullname="A. Banerjee" initials="A." surname="Banerjee"/>
    <author fullname="T. Reddy.K" initials="T." surname="Reddy.K"/>
    <author fullname="D. Schoinianakis" initials="D." surname="Schoinianakis"/>
    <author fullname="T. Hollebeek" initials="T." surname="Hollebeek"/>
    <author fullname="M. Ounsworth" initials="M." surname="Ounsworth"/>
    <date month="June" year="2026"/>
    <abstract>
      <t>The advent of a cryptographically relevant quantum computer (CRQC) would render state-of-the-art, traditional public key algorithms deployed today obsolete, as the mathematical assumptions underpinning their security would no longer hold. To address this, protocols and infrastructure must transition to post-quantum algorithms, which are designed to resist both traditional and quantum attacks. This document explains why engineers need to be aware of and understand post-quantum cryptography (PQC), and it details the impact of CRQCs on existing systems and the challenges involved in transitioning to post-quantum algorithms. Unlike previous cryptographic updates, this shift may require significant protocol redesign due to the unique properties of post-quantum algorithms.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9958"/>
  <seriesInfo name="DOI" value="10.17487/RFC9958"/>
</reference>
<reference anchor="RFC9846">
  <front>
    <title>The Transport Layer Security (TLS) Protocol Version 1.3</title>
    <author fullname="E. Rescorla" initials="E." surname="Rescorla"/>
    <date month="July" year="2026"/>
    <abstract>
      <t>This document specifies version 1.3 of the Transport Layer Security (TLS) protocol. TLS allows client/server applications to communicate over the Internet in a way that is designed to prevent eavesdropping, tampering, and message forgery.</t>
      <t>This document obsoletes RFC 8446, which specified TLS 1.3. This document obsoletes RFC 5246 (specifying TLS 1.2) and RFCs 5077, 6961, 7627, and 8422, all of which pertain to TLS 1.2 or earlier, and updates RFCs 5705 and 6066. This document also specifies new requirements for TLS 1.2 implementations.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9846"/>
  <seriesInfo name="DOI" value="10.17487/RFC9846"/>
</reference>
<reference anchor="RFC9147">
  <front>
    <title>The Datagram Transport Layer Security (DTLS) Protocol Version 1.3</title>
    <author fullname="E. Rescorla" initials="E." surname="Rescorla"/>
    <author fullname="H. Tschofenig" initials="H." surname="Tschofenig"/>
    <author fullname="N. Modadugu" initials="N." surname="Modadugu"/>
    <date month="April" year="2022"/>
    <abstract>
      <t>This document specifies version 1.3 of the Datagram Transport Layer Security (DTLS) protocol. DTLS 1.3 allows client/server applications to communicate over the Internet in a way that is designed to prevent eavesdropping, tampering, and message forgery.</t>
      <t>The DTLS 1.3 protocol is based on the Transport Layer Security (TLS) 1.3 protocol and provides equivalent security guarantees with the exception of order protection / non-replayability. Datagram semantics of the underlying transport are preserved by the DTLS protocol.</t>
      <t>This document obsoletes RFC 6347.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9147"/>
  <seriesInfo name="DOI" value="10.17487/RFC9147"/>
</reference>
<reference anchor="RFC10024">
  <front>
    <title>Post-Quantum Traditional (PQ/T) Hybrid Key Agreement Mechanisms for TLS 1.3</title>
    <author fullname="K. Kwiatkowski" initials="K." surname="Kwiatkowski"/>
    <author fullname="P. Kampanakis" initials="P." surname="Kampanakis"/>
    <author fullname="B. E. Westerbaan" initials="B. E." surname="Westerbaan"/>
    <author fullname="D. Stebila" initials="D." surname="Stebila"/>
    <date month="August" year="2026"/>
    <abstract>
      <t>This document defines three hybrid key agreement mechanisms for TLS 1.3 -- X25519MLKEM768, SecP256r1MLKEM768, and SecP384r1MLKEM1024 -- that combine the post-quantum ML-KEM (Module-Lattice-Based Key Encapsulation Mechanism) with an ECDHE (Ephemeral Elliptic Curve Diffie-Hellman) exchange.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="10024"/>
  <seriesInfo name="DOI" value="10.17487/RFC10024"/>
</reference>

<reference anchor="I-D.ietf-tls-mlkem">
   <front>
      <title>ML-KEM Post-Quantum Key Agreement for TLS 1.3</title>
      <author fullname="Deirdre Connolly" initials="D." surname="Connolly">
         <organization>Selkie Cryptography</organization>
      </author>
      <date day="16" month="September" year="2026"/>
      <abstract>
	 <t>   This memo defines ML-KEM-512, ML-KEM-768, and ML-KEM-1024 as
   NamedGroups and registers IANA values in the TLS Supported Groups
   registry for use in TLS 1.3 to achieve post-quantum (PQ) key
   establishment.

	 </t>
      </abstract>
   </front>
   <seriesInfo name="Internet-Draft" value="draft-ietf-tls-mlkem-11"/>

</reference>

<reference anchor="I-D.ietf-tls-mldsa">
   <front>
      <title>Use of ML-DSA in TLS 1.3</title>
      <author fullname="Tim Hollebeek" initials="T." surname="Hollebeek">
         <organization>DigiCert</organization>
      </author>
      <author fullname="Sophie Schmieg" initials="S." surname="Schmieg">
         <organization>Google</organization>
      </author>
      <author fullname="Bas Westerbaan" initials="B." surname="Westerbaan">
         <organization>Cloudflare</organization>
      </author>
      <date day="11" month="September" year="2026"/>
      <abstract>
	 <t>   This memo specifies how the post-quantum signature scheme ML-DSA
   (FIPS 204) is used for authentication in TLS 1.3.

	 </t>
      </abstract>
   </front>
   <seriesInfo name="Internet-Draft" value="draft-ietf-tls-mldsa-06"/>

</reference>
<reference anchor="RFC5649">
  <front>
    <title>Advanced Encryption Standard (AES) Key Wrap with Padding Algorithm</title>
    <author fullname="R. Housley" initials="R." surname="Housley"/>
    <author fullname="M. Dworkin" initials="M." surname="Dworkin"/>
    <date month="September" year="2009"/>
    <abstract>
      <t>This document specifies a padding convention for use with the AES Key Wrap algorithm specified in RFC 3394. This convention eliminates the requirement that the length of the key to be wrapped be a multiple of 64 bits, allowing a key of any practical length to be wrapped. This memo provides information for the Internet community.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="5649"/>
  <seriesInfo name="DOI" value="10.17487/RFC5649"/>
</reference>
<reference anchor="RFC9858">
  <front>
    <title>Additional Parameter Sets for HSS/LMS Hash-Based Signatures</title>
    <author fullname="S. Fluhrer" initials="S." surname="Fluhrer"/>
    <author fullname="Q. Dang" initials="Q." surname="Dang"/>
    <date month="October" year="2025"/>
    <abstract>
      <t>This document extends HSS/LMS (RFC 8554) by defining parameter sets that use alternative hash functions. These include hash functions that result in signatures with significantly smaller sizes than the signatures that use the RFC 8554 parameter sets and should have sufficient security.</t>
      <t>This document is a product of the Internet Research Task Force (IRTF). The IRTF publishes the results of Internet-related research and development activities. These results might not be suitable for deployment. This RFC represents the consensus of the Crypto Forum Research Group of the Internet Research Task Force (IRTF). Documents approved for publication by the IRSG are not candidates for any level of Internet Standard; see Section 2 of RFC 7841.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9858"/>
  <seriesInfo name="DOI" value="10.17487/RFC9858"/>
</reference>
<reference anchor="RFC9019">
  <front>
    <title>A Firmware Update Architecture for Internet of Things</title>
    <author fullname="B. Moran" initials="B." surname="Moran"/>
    <author fullname="H. Tschofenig" initials="H." surname="Tschofenig"/>
    <author fullname="D. Brown" initials="D." surname="Brown"/>
    <author fullname="M. Meriac" initials="M." surname="Meriac"/>
    <date month="April" year="2021"/>
    <abstract>
      <t>Vulnerabilities in Internet of Things (IoT) devices have raised the need for a reliable and secure firmware update mechanism suitable for devices with resource constraints. Incorporating such an update mechanism is a fundamental requirement for fixing vulnerabilities, but it also enables other important capabilities such as updating configuration settings and adding new functionality.</t>
      <t>In addition to the definition of terminology and an architecture, this document provides the motivation for the standardization of a manifest format as a transport-agnostic means for describing and protecting firmware updates.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9019"/>
  <seriesInfo name="DOI" value="10.17487/RFC9019"/>
</reference>

<reference anchor="I-D.ietf-suit-mti">
   <front>
      <title>Cryptographic Algorithms for Internet of Things (IoT) Devices</title>
      <author fullname="Brendan Moran" initials="B." surname="Moran">
         <organization>Arm Limited</organization>
      </author>
      <author fullname="Øyvind Rønningstad" initials="O." surname="Rønningstad">
         <organization>Nordic Semiconductor</organization>
      </author>
      <author fullname="Akira Tsukamoto" initials="A." surname="Tsukamoto">
         <organization>Openchip &amp; Software Technologies, S.L.</organization>
      </author>
      <date day="22" month="July" year="2025"/>
      <abstract>
	 <t>   The SUIT manifest, as defined in &quot;A Manifest Information Model for
   Firmware Updates in Internet of Things (IoT) Devices&quot; (RFC 9124),
   provides a flexible and extensible format for describing how firmware
   and software updates are to be fetched, verified, decrypted, and
   installed on resource-constrained devices.  To ensure the security of
   these update processes, the manifest relies on cryptographic
   algorithms for functions such as digital signature verification,
   integrity checking, and confidentiality.

   This document defines cryptographic algorithm profiles for use with
   the Software Updates for Internet of Things (SUIT) manifest.  These
   profiles specify sets of algorithms to promote interoperability
   across implementations.

   Given the diversity of IoT deployments and the evolving cryptographic
   landscape, algorithm agility is essential.  This document groups
   algorithms into named profiles to accommodate varying levels of
   device capabilities and security requirements.  These profiles
   support the use cases laid out in the SUIT architecture, published in
   &quot;A Firmware Update Architecture for Internet of Things&quot; (RFC 9019).

	 </t>
      </abstract>
   </front>
   <seriesInfo name="Internet-Draft" value="draft-ietf-suit-mti-23"/>

</reference>
<reference anchor="RFC10033">
  <front>
    <title>Hash-Based Signatures: State and Backup Management</title>
    <author fullname="T. Wiggers" initials="T." surname="Wiggers"/>
    <author fullname="K. Bashiri" initials="K." surname="Bashiri"/>
    <author fullname="S. Kölbl" initials="S." surname="Kölbl"/>
    <author fullname="J. Goodman" initials="J." surname="Goodman"/>
    <author fullname="S. Kousidis" initials="S." surname="Kousidis"/>
    <date month="September" year="2026"/>
    <abstract>
      <t>Stateful Hash-Based Signature Schemes (Stateful HBS) such as Leighton-Micali Signature (LMS), Hierarchical Signature System (HSS), eXtended Merkle Signature Scheme (XMSS), and XMSS^MT combine Merkle trees with One-Time Signatures (OTSs) to provide signatures that are resistant against attacks using large-scale quantum computers. Unlike conventional stateless digital signature schemes, Stateful HBS have a state to keep track of which OTS keys have been used, as double-signing with the same OTS key allows forgeries.</t>
      <t>This document provides guidance and catalogs security considerations for the operational and technical aspects of deploying systems that rely on Stateful HBS. Management of the state of the Stateful HBS, including any handling of redundant key material, is a sensitive topic. This document describes some approaches to handle the associated challenges. It also describes the challenges that need to be resolved before certain approaches should be considered.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="10033"/>
  <seriesInfo name="DOI" value="10.17487/RFC10033"/>
</reference>
<reference anchor="RFC9794">
  <front>
    <title>Terminology for Post-Quantum Traditional Hybrid Schemes</title>
    <author fullname="F. Driscoll" initials="F." surname="Driscoll"/>
    <author fullname="M. Parsons" initials="M." surname="Parsons"/>
    <author fullname="B. Hale" initials="B." surname="Hale"/>
    <date month="June" year="2025"/>
    <abstract>
      <t>One aspect of the transition to post-quantum algorithms in cryptographic protocols is the development of hybrid schemes that incorporate both post-quantum and traditional asymmetric algorithms. This document defines terminology for such schemes. It is intended to be used as a reference and, hopefully, to ensure consistency and clarity across different protocols, standards, and organisations.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9794"/>
  <seriesInfo name="DOI" value="10.17487/RFC9794"/>
</reference>



    </references>

</references>



  </back>

<!-- ##markdown-source:
H4sIAAAAAAAAA819+XPb2Jng7/grMO7ajZSQlEjd7spu1LLc1thuK6I6nZ3N
rgckH0XEIMDggZLZ7t6/fb/zHSBou1M7tdOVcmwQeOd3n/1+P2nypjDP02eX
s2zV5OVDelWVtqmzvDSz9IV5zKfGpvOqTm8r2/T/vM7KZr1Mr+rNqqke6my1
2DxLssmkNo9fNcifr54l06wxD1W9eZ7m5bxKklk1LbMlLGJWZ/Omn5tm3l/9
Y52v4M9pf2GX/akfrl/A17ZJ7HqyzK3Nq7LZrODbm+v7l0m5Xk5M/TyZwTvP
E/zKlHZtn6dNvTYJrPAoyWqTwUrHZrqu8wYW/1TVHx7qar2Cp7d//vHm9lny
wWzg6ex5kvRxxfDnTXUPf95fX8Ofr8Zv4c87eJI8mnINE6WpDkDLfgYPeFHP
foLB8Ty+x9/x+TLLC3pv+ifc56CqH/BxVk8X8HjRNCv7/OAA38JH+aMZ6GsH
+OBgUldP1hzA9wfPksQ2WTl7nxVVCZNtjE1W+fP0fzbVtJfaqm5qM7fwt81S
/tLU+bTppdNquTRlA0/g5JfZagUr/F9Jkq2bRQWHl/ZhQWk6XxcFX8t9Xq+X
WWHsU1and2Y229ALsKaszH/OGriD5+kP1Yc8o+dTONbn6XdZ+QALqw09q80D
vfU6q8usyT7Im9W6bBAObsqZfGzkgD4MmmDW9zXO+qcS5xjA8mHrW6t8kZXp
T7CTjrVd5bDzj/SDgmrwyK3ixzJvAF7HDUJYWs3Ty6WBE4sXNsvKJ5jlTw/4
711r+c6U6TgrGlN3rObH1+kP9NesSK82AK+pAmN6BdciJ6bzTUw5sEd/Kqd2
OnioHgfrD8+253td5zZ9/ZRnzQcAjw95x6y3fx4vclPM4lOG7/6ULavyYbKB
Pct2yqpewlePBNkvb27Ho8MjON93N4Ph4eD0cHR+8MPN+H6AvwzgJ/fS8e6X
jt1LJ7tfOkkSJAjB5OPb88PD/ujwvOOj8e1AfnRjj3aPPYKXbsbvhhdnF4fP
6QyU8N3olFWZWrkIwBG8GP9PQLN0VeeP2XQD/181Zkrv9/3V1QZQvzaEWETq
pp5E5tN0Wc3WAMvPeOqsfjDN81TR/enpaZDbirCcUDqrZwfno+PR0WDRLAsG
XMVO+q+Pt/scd0QPiN6lL82kXmf1JoXtnsDz725+eBHv9cdyksPok8Kkr81y
uUnH08Uyn8FS3r7pv75+mwIYlSZvFgCUby/f9HGE/uv+1X0KMBE+uX3dvRMD
h1Q2gzyb1rQbWMnxwQnsYzWbd24jL4E6jwe8EPMQ7OYShipwKwg7r/58Jd/o
Vl5lyyVSVuBINu9fbaYFnPIevLf/TF5sLY14yT+mtCygewfw9/d2Zab5HFAc
b9O+x3N7f3j+fjRy620tWFf8fTap4NpT06RZMeAfZdnrh7Vt9ApeZgVwodbS
5SH8CBz1ZQXwA8f9JmsaYJP9SWaBBl1Vy1U2bdJx/gD0cl0jPXqEt364v/tx
x/7mNGrfwhcDRCN5sGMnso/b/uUghSX8Y22i5/86SF9V87ltTF7GHwzS1zkw
olJImz7/yyB9s1nD4hfm0X7YRL/dD0B4qMvWSPi0Rk7eeniXTytrTfz8e4AQ
kxetSX8apD8tgHBFD/9tkP7bIhM+IJfybtpUSGjhVg6RrACRzZb98e2rmx+u
xq3b4R8RtOT3PxA6X4NkMZsF4sza4juAKun1x2y5ApQCjnF/+9buuJ8O1Bge
DA/PdsGabOdukP6Qm5mpF9mDKduXdFc1i/azn7JiZrpBcohUobJ349GotesX
eQFon4Noh5t9a5Ygn3VJcV+/udHBUYD3O/AIlgvr2X54Z0rTfgyQOl7Vpvxg
Chtu74WZmqXcLpL5u+srpGWfYRmjM3gNwPXwwr00PDw8O7g4O+8f9U+PR/3h
4dHpaf/s/REisTub1pld3f2P8f3lm3Hfv5BeFg9IGBZLWGtIW4iDjNerFYhl
CDcvqukaeQXznb2/ALPBvxwNhrvpVx9YCrCHwjIN00kP4Bgy/89+RNT6IHiW
s6M+Xj4A//nngQ0I8XdZHj16M0hfrKeZjR5eIxUomp/buPvGrCqAg6+mDLdE
+J+yifkKbH8BTxuzKCLYDnkeQvft66vxcBhzPXyWfjMcxleS+jMfdTMz4BJw
1MBebL9agRiGp776MLXDofwfHfXBIwxwUNmD3Wz63eX4Zhywtn9dlwYXfArP
EDBViLhixSg3trWBXarXcy9/7F0/ZsWad3YFT0Byzfa79zW19XRQ5rZBgfIA
xJm/gzhjD1Y4yz94ln4gvWx2/9JXaUXEzAPjFgHv8SIOVIrq73X8ur91anJm
eC7hkWWlXPPwjIhYM7xooSMTrf71HG44B9RKX+UPiz7cOVCvGyTQHt+AUr8m
yRsPCzDSfOy/Pf5q0ja8ODg+v/gSaXuDpK2pt6jb2wESuNdZCXqEnS4UyHfi
BO//bUZ7v4Bn39dmdOgI1/HpyfnpQQMD2cEjIsEgHw6GfZKaxqA/ARbFjM7x
Of4RkGKRg3bxBzj2HjD90oC4e5OOF/W6/EDszcsgv0HmOxyef1Hme1msF7Xb
vzz/8wA1uYfOq6dN3VxfX58fjgbDy7uWIA8/oPJGAEls7E01BR0LKe9bAzex
qoBEgpZ4CQw+/cE0qPlbleGNcLj0ZgZgQpaBrs0qwFvQy42h/eJfDmRFB6cX
FydMCaxBREZJTHeOC4R1yqtAkofnzz4j28PbwSng2/DP1z/dIOGgd3Xj70oR
Q4AaoQL7A1lBEMhvAMeE/+Qlyvgvxpd0n8CCgFeRmFmYxqSXU9KCkTPBZzja
nfm7KDkgBcxyGoRW+xXXf3owOj5B4tbJZF4P2qrqlpB26nXK96aGLWTRhvEn
eO0YRrwFXQzWDdf842pGevveNX2wHy32WTftmwOTAcK3nlj468rC2o/hL6CY
E+2nZ3BNx/2VztJf8yyDj4X9+Ky9wxbpUmJfbHRPcntX715c//+5wXTvqpqZ
HWczq3K6QaAqJ6Pz4cHPpqxm1WB0fHh4cnR02LHdr7/PpN/vp9kExclpkyT3
C9A0ZyIDoUb9mM/g7h7W+SwrAQlh0QBU5gH3DjvayQDTvVvQ+PDdKgECBboU
qFCBtTCdsdjaS+16ukgzi5Y8UGZxMqQLINCjOAAvfjDIzaoapOx0AQj+lNVm
AOs01qQWxC6zRJMQgEEKsgAsy6RPIG+JTS0t8mXeyD3B2mFDMCkpCKvqyQBh
vbt8S1aEZF6AJJQuiVexXWEJpN08wsATk05ACzTAxOgrMxuksAB/Tma5WoA4
8jMsHi+4rljp0OUmymlxn/jCBGUXooSWSRwvHVeJB+KE0bkRHVMPyQLPFE0U
ziUBxUO+S5sqBd0IdvuzSVcoQFnEjFROrpcax31BB5sVAo5mtQDuWwOWwnCW
tw0LTEDFLKpsJm/FFpMmsx8I6Ivqqa93m5ryMa+rkswsA8AOUL9tlZqPKzQ1
0q6THDDCid0wLEAI3sk8r5d4SimjMFzBFJaY2yVNQjsPAUfufMCQu8xnMxA8
k2/SmxI4yWxNiIVwbFL4orS5nk4oK4U7UlCF3/GgEYNREoWDgmUAIy4f2ET+
lWD8ecAlSIOVmro0DemlwOMfAPsB+vd1sEEi4C3/Rps1jFkAuVIQB9UBELgT
xpPdMJ5uwziecLU0KSgSiHl1B6RfziryHSR4X5mqUf5ywptHO9UsR1BbF006
Wxs8ebj7HBYLxBx4NBJr2EiBNK4mIGa8gcX04tX00gXIifDSFEjpulHL7Mws
4V2AMVSCzT/WMG0Bm0G0WuYPDhtotWL7szBgbebrIl0VIN6ReQBgs7SIeoKB
eBgeR2BdgP4lXB3hN1KUvEzCi4eRGrRP8m2FRHNWwYRlBbhHWg3bJatpVQAd
WWSPOQARsCN/krCCQvgBmSeDSUqVhXB1hckeCZNgMj5V2F8BojsTap4iNpyB
xIrHiUeBEu0STwLNVYYuvomWDUsDyrGeoLmOicE0B+qAMjjcK5KqwORV8sU+
VfADQCKh8zID1GnWM6OX2wAay+WbosgBiKZ9OGuYhjiiqVdZDUPvAVFLP32C
e+gTyiA4/Prr/iAF5gt0CnbPBB3F4nReZ8uIa6BsSYLbyeAYHptyxtYfwGYA
5yafA1uzHhDg+wevboBoVFmLN71xIz8BjVvA8UxBGEWcQLdWOQUIQ1r7sKH7
WS9X/D0CzUegHAhBcCAFjNZLcVuIAFkNgA7nvARcyPuLagWYB+inlwqAc7VN
SQgHmZ/NDBDPDfwE+5wWOeMXHHHOvLdaNxO0IeCKSgYgmHyybhiHiP6iqRKR
ytRopES+RFdp0LQORLtgORxEB5SiRP1GaRywANZ3CauBo7Ow/h6sBgEA7xr5
YonKG+wYTyNHWoQwBv/OZjP4AMGsnOczls3IUD+p4A7dkpn0yN+j6ZksOVhL
Yd3eMrC2DLmCs5OqgoW1+IdwMT7OFImZ5duG875pAX0vfbZNzJ8hEaOZMkS8
FEQ+wLMiSwFi2QqHcA+PiMoTqSKHZIS4KMj0mB/M6Qms+tOnf7npvyC/YT8H
fAUZ9mw0Op/kAO4oUMSSV77M6hwAiKVBm14B7bbp0MMJHt0EsHkgPx26n1gs
YDI7ZV0RWAAfrDAkvD/Dyhj+HJB1JRCA+oaP8qlaF7OACendi5DAfNwJMAQF
ARvscrcM2ufV6ZNhvCfuBq8QE3PiDKw74HPTbJVNCcwYexFCQWgzahqGwweC
0CDxzBrCMKTPsPEl0N5HoopuLbC2T5+QEHn6/+uvabYC2cUg7OclXkr15eXT
YooO0taDR4AnOTO0vqIQHx//DPDfBzbt/wHHO8dlwHV8+qQQL+ui5RB+bYEg
y7FynAjOsKaKzv9Lgj5+OAHUgZ+AiLg4gUj4b8sDrVNRcQalwRr+bp4siixo
S8tlOOSzXkQNhdGA/YqXD04A/YGkhQD5XYppCxAkRxlzxgR4p0gaCewinKI7
AVYSTxwI1XnpOKuXwe/fjGFFRQUnQADagLxa5iCIEOjL0oTpOjIW4AcLE1+G
Hzq3nM4VuJoga672MmG6iLwdZ04n1o1UcMvTOp8YRz14FjqRaCrQfGZVLTij
9274GHSweV0tkaTWM+Z6CPVwCAtTrAi9YDakmrA0wUzE1AlaxHNZZHiVdIDC
qnVoFh0J+2pkmS9hZaV5EnYE0wVyHougdkH0aoIwDOsG0ZGWpBK4l7ZgOwsV
UyaG9K28QVKTAfutn9CYFH05RaGChbSKvkQJgaQshC0v0aJ44mUlpqAcbcKS
t+o+Ae50y0CC7o4Y7OPhAjPAuRn0FsZR42yK4SMVqVDOLIK3AmJNg1I0+t7U
XYkfCocs8rmBKwD8Cw6ObRYCCE/AnmFHP+MOQioszvctKsznEaDFtqyJ8PGE
HIZW4k6Y4BBOdQbiWP7RzLych+swH0GpxYlYMOLbJgYG8NdH+Ns43GENUM87
IhfRuoLTFe0HGT9KQSgJ2wUyj9hEzfIQWRMu7946KzVIaMiElzmIlICQoIui
EVdgA0hZsUGVH25eTEW1MwJZdFIS9MEnFdDIT5/EHIXkn+45I+NG2RfCBroM
HB0SwtbaskeMT8IoAtpaYNVCCeNy2qwBJOYAtTnh4AZPErCcNoMXwbKG021Y
YnBQ8Qii3Ix4OpLGZouJzOEvlrSDNnh7bwR8PNnQXD+OfbTNDRAwUB0aspqo
pZhR5x6vqyoqkLz30H63T6oIazl+sJu7+5fp1cu779WEkYlCM61quLBVxWoB
xqLt0pWA/3B8hb95gLu7l1dWhI2HWuC8WdTV+oGxj0Z0tmfFhoH3aBWonOYN
i1Vs/eF7xpH8EZEE+ZaAt68RB98RLL4GenBdAm4B2PEtv1UDSbrH0SH7cNcS
DgT3vGugF/lD3sBhO7dB4BPdY6j0Ax3zQBRyVaBA/yqzi68ZaPzmVTzSCY6E
0XlAwCl0bhp9OiZzzsEbA7p+U5X9t/h7Hryw92o8Pnjzdowj/gtcx/nJyTEi
hpiqnCBu/iq8662pPxQmnIJU33Tvr2/HfpSjiyEt7NIJYrGFKIBepAtGLcAo
apkYoAHX5mucKRQ/iZ+iWIUcE44LkRKJGGBI2uL+Tk4a5yR5+ZlxIKbbqBAg
qav81FZc1PCYLJVwHutVui7DtQHHKEkoA20WsMtJd7lVkW8muA/LYokfFUje
TgtBaBb5KjIIfxamkfxxXE0gDbExAkGEfgHau8iIcdeNsNBYC7TMpBlI/Sl/
QBreoAxsMqGgdffJvqqezCOawnIyBhS5rv37bA1InQVU2A3r7Vhw6o7QMkWZ
NiB1EwRWjiym2QPMChJzmyg3DWhhsIhLkoIRg6KDJa73lJGSkzVMh4EfIyl8
goOENQ3kHF8BVf30Cf6EA8OPYC0zDUwCMoCxpaiUIZ1xtwCS3yxH0oPkngR0
t5MiimyCAVB4WxBpy0kmbAE90QEnhjfuTuBLohXqt2QKEIgeFvXXT59aXk/Y
hOhpX8YmvBRCHeQPMj6wdkCLDHZByFV1I9VnmBS/0i8AMoo0wwjpICQk1spA
ykK6jt7OCBiQzzmrIdpndyxCWG+HJTCPTAQsV+picfXfIANI33pbZF5+Log7
/fRNS3dlScipwrFSgNBQlSRIqW2lU5uojdp/CBjY/lKoktYalEVgtmHtNFTD
TCRLbpukY3OvanMsSVkTDYLrmq9rOjTQ5qamnhAvEKlAjXgz1jPVztxtUN6W
yuH035VoVW7yBxeeoLJvtBfLhBkBsiLzwMKQqyZSMYhMrosiAbTKiFNR4CoQ
jg/o2hfXEH2GzqGsZpMnfE5rzhBg4QMWmbvGSLIATgchK0dPesMhNaeDIyc+
iW7npv2vH/NvVf3F21/XSBCZ6g4ADL+HC52SssSbZfUA2S/zNZJVo9ff5+Tr
wFDMe1TtZuupcUcYaW3dmNML98dumBuygBZxVMActcN07wYe3ryw+3RiFGQA
lzEOXDr8URJ99EY/+vQpCGEIRYxqJaASWBPDhcmhJHIo4qdDGAfZJpuRo03v
i00bFGJTIHp/A3eDQotH8E/f4BPBW89XASbYrusFFseEUf3LBXuRjdFDuo1A
MOyFENEL5TPRMyU+79dfn+O6ZGFjvqwkGSMD/YyUhACQMRzBchSKRIOPbnHP
DB4GPY1spigQlj4RxdFuz6dHyrpgACFQOEgPpBXcJB1JzHBVbKpwKDIoNFle
fh4no9tEXEbdAUUWAVULmq1hXa5972i8Roxg5Jw5uwVPRrqzGFurWtbChBbY
dwZykNqD+BPxEzsz6X1AEjKKH2P5WV+8dUx272p8u79lcdbQerxxWhnz1KZj
I34PerggBSwxQBXxPZubhzXwaNyTmh5AZjIJs08A8SD8niyNwfgDgqXQNOM4
iOCMHELWabRSQ5dYLOBM6Qad2QA1QmA3+HcRZppKl+OksnXJIRIkZGRTUtMS
DIrAIwZowZEnpnkyhq+dZFk+e/TWlZ0kFw75sSILMhAukMUwNDsJR0HgUQFh
umkbMsU/92VLOEB00sVf2NPxWa7Aks6S/KtkcoHLITGDR12RMcHt2CFERLAa
R7tj1ok67QIpnLAKfHdtYc9fPFm3A3FAdK4clYJZjtJ2K3agDA+RWCe+Tgeh
hlsALxvSymDhjMc5BfInBFq59Vt5zDA8TCwj5GcrYxcJxuXXRNt5djHoxKbj
XsKI5v04DlJWVbEpqyXysMAq7KPWJcYI1gj3Deie3mMcA9lj9n64v99nY6Fy
9LPBCaJHQNr3E4QKODsMLegh4YguDZbi/dz4Sya2c5Y9I/IuTuNBekPcGWDU
Nr1EBfCWsdw72wmcMRKDUEj08aY2bDEkdyiG5f/EwkHe/JyCnNW/B6UnCfT+
n97dj/+w35rFxruP9n5CFlJQFYhdNmEsB4qqiD3tkyCDt83JDF4m5Ad2knJg
Zew5pWNbKyK52V8zSkTOIxk6LwfO+4guw5CeLaqCTHCC2E5/cr74b8XMH5qy
IiIE8F9WT2Viq8KwO4jwWJfBjluWZ9oeDqfROUuAvP6336EAM8crx7uEK0lC
zinv8hbawRg7IGQKsOXDIxgBZYQkoopeXQeKMTUlHG1lA4OxHJwKGZbUejiA
aOJVltek0ZpEAkm6loOyAmyzWLM4gfFVSrdZ44H/m+bW6YVMF1okKGlpi8iQ
nE0wEMlE8vn0iaPhMcqBCOVP5Ch8XBe4NuceKfIPvN2d6WHPQDmx+BNhGtkZ
YHBMBQOhDog+85XGJGQoyzFuiwI1ynxFxkT4qovwknYWhfWgHxnfVpYLbHb6
wcwSz/P5pK1Rc0fghM6sXS/JG5PN0BKBWjTZe5D3OfFBjOZEiJOIJeC1cSpg
DBlZeznK89Ea6FcvbE83mLB5CSHJebT9dyQjsYvLnRIGJYjAtMS1L6tG+QyF
8DFLS3YKmNFGcgZ+ZmuIqzlTEtUy4ZB5+vbFqk+jWwyxO+RPZisZRuz0n1As
jsSL+bqcMm2KQj3yeaSK4nqd4Jo1bTRi5xGauLMSX52gZud4AACJW1F0EHui
8NZ4nzbwkHZxLMBQ8loUBpSEOGfAJiJVitNKYhpo3WUfV/KoSLVhcZzxmk4a
d+htJYyMlz7wD8AkR2MRfQb0SsIySRgUX1QsDFLsA0cohgaTUHYj5Qbz4wxC
MHCPmtTK5aqxYluCWQhZxBPWbS5oCKZzm2QIKoAbADMqZOGxeqRRyw4r4rCC
PR9Fuu903PYsCYntZDvLAX/dUUzMNFsL/9IxdSKCQtk0OeyieDohq8lEJDWV
d8ITjIBXAyh2yoehuxBk1k3CYENBi5aivYnOEL6JUgS3A/MWrVl5+WpC80pt
jwApQSsScpkscKw+ZCukQXQEBLoMYXBkPggKP+5t75FmY2OGyI6fPv33u5dX
F+fnQ1HGRTeW5xdHqKQjKGN9gJkgI0MiB6AxXOrBDRKGLoXZ1rEhpUWdiGNG
Il2ijaXMPom7WQqDA9wwwGRg5yqCnQ9GLIQFW6hq/3PwI+9DQmkGbGTwWUFo
53zhRHjhi4HKEGs+rCx+QXHqiSF7CpQILw3NXUkgQJAwtmLz8Xw3VIYWXiQF
hoLpWNydeZ6U1CAboOWswKzz/m4gR9SADwBeozV7CggAFNFAClfNAMPdKz01
blJRj9sf2WUgVwgqHdZ3kMhFBpKt4MWei8gg1R9YIBwDpttPaYGtMOAgmCcQ
yRInkllD4hQz1qmYKej4xcZP4rcnCM5eGkJcEp6S45B7eekUORVtEQS2OUFs
+EnE8LNHZpZ9+IYNWDAsXWWf6CKK67VKrn0V5r0khBT9scpnIoIHbClxKiPL
hUW2iaAotlwG0SxkCiEgiK7U68wtZVacyj4YQnwuLRMrJwgz4aBw3SXWKkmU
x9sgcKVN5dTK46OLBMFQZCAbMUKJs9lIFHoSWK5c0JVDycbGZhu4rXmg5RIP
IpQGVDVFT94kGSaCA+HuxHCFBlBcv6gaC1P4MMbZunaGerqvbvzqkYaMyh4a
c1umO1Qct81U8KVAC53gE/BCgLaZSdalEzGAipBUywYwpjwlxqSVGBjERsBA
gg1NIZ3sTTAnS9jTxXF4cM+g5a01PA3Rvo/FdHIszcG0deq1EDwYl0WA9+hh
XTWuXYfkCLRjL2zDw6O7lVUCBFqC2u/zR5E7PIfEuZ2Jbp7lBYeOi72nhNuf
9zH4iJGG1yisSWxJFglIgiq3tTIoGvaKrBQSNf2wXtGC5A4229kq1i06xMdB
Z+i13Lsk4GiUMUFqysGKfZnT56j0SIMBwYZR3qlDDlmBBOOCzCwKm6LkA779
rAnsvkDROWqspjAB2CsZub2jRhaAqeXixQ81l24jJ7FRLyZ4q2ZboqQtyoIR
wMh8WDt5UrfnUJjOB0+iBB2OaCcBHFFuRlOyLfKxauhA5rcLNwBiNgY4NZje
QOo4CLI6we2f/atOTxR0YbGGhNQfbbQi/7aUUGJLOEsgJ+doCX9DPw8DkwoR
+svrcX84Otffj9yzixEbT/j5iT4fnZx+K5FpVGgCXxIK3B8ejo5TJsgg4fXP
z1jpkRH2Axrjr5Cpo4zM6h5mpiCXfKrgWFHjmFK8Q10hT6NF+3uNGE7lGA3C
RdbhXGMvzzfpC76ue72udyjdZJp8eyW3++kbvta+XuuvFE7fZZ3ZnorYB3rO
KLJKIcCDDqlNK4NUMZDOyLuiq/JoFJNI9qQF+jQMVTKX63YlcPhj1ujkbA1u
Laiko2MvAgHtck0O/FaqBIAjydBAzai+AibNxIkPkumAwbP4r8BHIqCsG6xJ
E+Xd9IREzPpN1TflLPgMo0h4LWhY8auBr+/fjNPh4MhJ4senIGz73BCkFpi3
Y3xRlMwHJPA+BY8sASUwAjRDfZu+aI08PD5DGZ8DTdQPFHzBZ4wUCs5+yTsU
pXYWiYNRYJRte0IC+yvgAsdGdd49ucaQywI7QTYOdwmiH8f6OPhBWhnm5VDa
DpteyMQGnOoRw93L6glzXQh22LX9LN179cOLN/tih3lOujn9lbQAoCEzKzK+
IjLMipIuDD7LNiIe0ogkFlQU5ZU+i+CTLlTzuRKXp8gKLa7h6o4Saq0P8iRd
U+xGhk0ezQJV1V4XeVYZz6GhevIy1nE+SqAZ4YbLB4jdAUJemejtvdgXwOip
Xbvk4OFFbh796IvNpM5n8SSwrwmz+eurF6+uXeSAV3qHh0BAWZWUcdiiiiX1
9M1oSPjMZdc0he0viw8Y1TvgHGXdQvQJD4aO9XWomLQ9h3j75GRqYb+3d03q
6oNBD0iDhBklsQIOaaGSHl043h/zdXeBvQiUpsQQQAQHtAL9oiGLqVIW9bDp
bbpbZAunbeW3tnOq2O4cxGd0HNfMZsgb1ZrmXGfFRt0cA+EYLBIiBbl2VLlb
OAS24Qh3nzkV8w31xiw74z4CthIkSXtOh3E4nXJdj+CkxtwMu1kuDcYdpYFA
45Qer0wBuyXTw0+Ah3xCt7h3GGGPIfHk9PiCPEueZa+FcsjIkTczJFETUsRl
4YM0HI91/JVMRWkG5axoUbYnCocoWEYSS2zmHWhwDOdpBRSnCZTDXZ5qwmvn
fJljMiX5cOAp5eFhzFCSvFPxnwOXI2GLZE61EvnjZaevmvz8WTPBQfYYCTmf
8cfj8QGbDlxmDG8u2uGuXctPINGt2oqNPLhxT3w1gqsTcni51RRmShFzSOsL
FNxuv7x3jKuHRYJ9OCXUSfom9AIh4AREOJCGtYQJxQVdu6ykOAIwSby25SPc
MLpNQ9hCGNqdVgW7AxigKPmmSiYuq0pukm7OFy5oBS5wIg5OxnGj8STkeiNa
56PJWA4tgzfZxsvpRRx/mOQAVLMcY843YUxB4G0Q+CtRHJ5SPMogfckhgEhT
hNgShC7Is83SAFrKXXQLQbVbRogk4ZFwQAzW8JhpaHbbkgfKA3FJhpHky+ZI
gN3qUdzTetxW01QocqLHAOViMaqqoToucLdJ+27Tz8Cm2O450a20GkSNAg6g
ng9n5eOxidSMgE1vKj7ivPb5YHA5Uaw80oZ27QgNy2FTV5250Atk8amHlsjy
xHouz5KIQ52yc7yBLGLZGnYhvuwV6s4YaIzckpO5g2xo9beCmLI/SF7gXZLL
M1w4h3zLeVLiF0WyYYaSVdylc4sOPnR1carw1uRqX+XsWW81yqVOhWkhlxQx
JWfLrtxwZ5/T9Muvwm2Kw/9eA/K6MLUq+xwitANLNUoL3VRVzYEOX6CNGBF+
LWnmDqt5MLSlEXWk03bGhSwnKd1TDGU1u6zk4idkDIXp6IK3kdRP06YIavjD
j2/RDymeKrpTvVJMA8Bsqyas2ACKixEdm0pdc6T2O8ZoCppllH7p0BeOKiqW
005PQVFJU9EA09RR46MK1HeJShHX9thVzwTDQDu5Vc56SrYkPykd72ZFXEMo
UKBa3OyIx2WzJvuVMpaX2TDxsEbVXl2QjxV6zguXgrd3d/kWRCgnJ5DRWqEi
MKtbJ+jusEqyboyuXTeF2tUE2ymiPJxLDa1MpjQqRoPNQV1/ULMxSjjd4V0w
XMnxGAFhaye0eKPfHUjYKJ5c0acuMiqsLoMZ+bJXdUdLvRQn330xEiw8FqBN
df6RZT/rT6ojz7CoqpULp7eNWVkNTnNHRcmsYaTaIwyCycBqznB3h5UDjXVh
V6JdkHjPV9/3MVagxbeVfKnm0yqaoc4ncTQE4IfmAyJrQVmAIO1xu75zeERU
zCmfh+peJyDFSvaNFiHiojS09IcFJkhiivBW8DHJYA4gKDPr45T8q26VlPgg
B9ZglrQPbyCyWq2bKGd1sCP62tV0YcXby2+xWCooGIoSn49o1CgqdJ6j5REw
6hEu/zlbSubPKEYyQxmcWVxGiTgSiOwWsCY3yBytuhkZEmiDPqr+md0aaKWc
UTLBfALYAEP3TYG3iRn0diFaExJqlxCUbegneg4XihLx1kQu70fKS8CuPTQg
MMmg7a06KbOVsS72LN0epp7AzktnkXYkwnMP8qw9s66c8RJV+voZQxmbJrc1
Yqe5ab43Rg2KKwW1lWAmLDkj9JyNTVyIBOTltbUaGB6XWqZUyC5vKoU1o63P
e/hITFXueHX7Y1BfqgubGLichTxIv1ZiWWQ/bzzdUsjFu5CpGjErWwJdr/ME
fENjRZSAxTFXZTeeI97IbDWW6FsaFwaM8Kb1blTvz4pMlABrgnhvnzPebKdQ
w7F+p1EWWwkPYfoEetmlNoGwLQmgI9k7vX+qdqaxOwftRByAElGGOXMhPXY0
5muqT3DOVwwzeFF9d1Eub2RVm76EGxMcge4q4k+c4TNP44RkEXzkhDpOB/Px
13nRuMBUtDNL1mLqwlI9RZkBryqRtwqkesGGAQmjGjnHGyg84s+W+1WY6KUb
MrO2mubEgzq4oSR2SFol5q9nAGm5RaET443E/OzaGAjhJQ7epH9ML//2e5v+
ITUaonjJtmyJ/JXFRN7pTLRmWh8wIzYxMAXeWh/xLjlMNg/Hy4EnmGGKeXq6
miEsx47EVAj/ojTckZTHEKsyCs8dh+Fq6yAr161Ggjg63WuxnsWiDzthTOFB
hSt40RFoPRMng0SbFTo+OjllPwtZUl2giWV4rrB87UtxZa7LFYeL1gZA1/pE
XYpTRQNfgU4APrjf2XR01J9gIAsOtCZDhU2P+9iYI5wJi3zMrM+xYHiGz4ej
7c9HOz7/NpWlRYEGnrF3ZToht/JKSrwnBgE25YiwQBKBrCUP/De1eQBxhnJ8
MZmH2XmtEgiXzwEKybKlXq+AaHA7PW/2nlVBtDQBQ5gRLOGsWBoFdULxStHZ
8xmJI/Xs9DyKqHV54PFwPReZiG6fYKUs5ftzJkPZAwZLaHIZNycKABjppKFk
e06/+1bEKIBJSubLfEnXaVbXzuW1LtlKQRRB8iA5ZyxC6MsgUoe2x3kubsVH
H48EAcn+EaAaJg3DpAjrEYzDmgBHsVMMfYIxCZnalTNnGjz62+/hf/Ax/AHo
ftw7PTyXr/bIp/wxX7Lp7Xhwkr7+bt/F1XiyiB4T2VMHBQgIEtqdKKQ7nyBd
jHMxkDtFNi+/Wx5LOAduIVzysHdydNq95CEveZC+MiTx4NKzLwjrvVAd9sxC
I/sQPfiQ/oCr4Mn/mF70RsMdi7jgJdwrToUhfJpihy4nH4nFQjeV9BO0Uj6I
IkFRaO0bhSYBfYSYJYpPktmw3Oa3PbW9e6zYSU2ldhHX/OL36TJCvEHT+ApB
/Vj2HiLliKmrC3bwDK0F2+cfz3pO3+XEl8uOokbCNrAYkluoHfaQFRHnO8ST
gpHO6Z/nnnNgOIcRD5yYQBHXYeK//f5MIOkYLvHkrHd0rDvBA2aWn7l08tFR
8PboqHdyMgrexjU6gmerL0KaBBS4MJczhBTcw7YVoKdeo4DK6JSg/fRzLTzt
D0bkaCx/R7v4GYV5stMAdYUTBbXBhRp5TrFLT/Y6tn+3A75acn2ULB4lPvVC
JVQEMYuV2kjB6zKrB1UJtFafRM9xSNobVBiuncJA2aYicb4LZeR7lZHTT9+0
ZFdmPi3Nw8nUJIaVLYGbArUiMsa6Q2y/RwWQAkZVYFJ9LQh3ZYVtW+jcUtwG
VBlJUshRzGaCopHfW0mbPFKvvS+fJ8UVV1GJwgn68GF/Thk67YKQrpARexEk
4khj3USMwNqq6lRA82jRkuC8fFAbT1VtjgaurDTV2qIfIdMMdybZrUBhKfrB
54I+CPFwCti5+qOt8FzOc8LJ+8JRxKqmzvvmjyjuMq32F2C3HFZRGg1OjL4K
tOL0q3Wzolo5HF+b7v313Utf9miEvvSlyUqHeZRcKHVDsIedcgRNWoDrZ3HJ
B5q7oNExB/OqS1a2ZCUdkcFI1gsi/gxzmDSkHzWSktR2jv1bWbOeVX15za/+
9u4lZqO9iUFnntdY0YZrLnCBc2HWrfjxoGyEhmJROgzpxKDGISWhWoPkvPMQ
2XCiIkwSnE20x71/t3uH+/++j7Zn+cjVsMOmi06rolGmVbFe0vVf9oJpKPJJ
JsC78PHyQGbXy3XhFsM2qEbqWTa05hKuPRgMFjTc/3cxr2LIvw1xxuEQfcXr
QfNxkxdbOGKZ0E7QZC6fk0Mzkg97W3JgQzWKTobKkva+JBHuc22gkkrUmPCo
4WM3zn6v/etl+KtTiuVqmnQPmCT+Hklm+47kTQyHZ0d0DdGAU8ayCRo6RyS4
kSuzaiRRW1OSt6Wrr5LrVAJ7pNrqJNGj5VfCqKsS2OpUaJ0vKWXqMCypRUEJ
5D8Ys2rlc0QCCsbrsvZGCr5yh4D7ORo4l3raRPYIb2PJgOV2rXgdBAxmzbcp
14qkbjPOBiPNw+DfOLSEEcM86Nsgd+JPVEs32hXTkpagwjwfhIoCz7cIU7Ja
Z/7IbZoG6SXI+lu121sHyGm26vnszssPCgoM0h8qjJBA9oAVbsVl7tQqd3g4
rlYM9cwxANINhvEEZU98qToQHwu00QyH+2nzxBkKYYZ2VGpVQfoJ69Rp4CYb
AJ3PSy7ZwV0521WAQIpSdZWnfMq0yDcGXvlirWSAo+5KamO79Wa3rxOBQjtd
kkR5ag5Cw+rmLg2yFxoqyQsmt0AxoCg0I7xX80RnCPI8/uty/W26t8RoAyr5
460Tj2afvUfxLjVx7FTzyoKChPc6IH/nohTRLCAtTDQrN9gsiWQqZOCaaImE
JrSqJF6VUC7HloFIoYjxpUgL6dUhcbFJcPPzwnzUgpsTLNSHoT+yFBFvdknl
rs4us5Qk3BV67oJokY4L/KKLnvR4J1bCXoPDZdtQqcV3XaRsd2CLQX9cmTUt
qTchi+Myb0LBV4EhIIKDKH8wUbuzDZMrrj9ynhOusR/T7Ub2ER5NstfxwQCQ
Bl/ajwsfkEdGS8js2CEVK5YURHcPKL3LbvZwGhrXShwpg5oWD+4KRg9OoHu5
GCWwP0i/28ShH5/ZGdNh2QmsOdk1bJxBv2PT2xeO9jqs+k5pOUFeJAYBsK1n
Ftbg8FlLlA6Ejd8eUSIyjoIHiidy6cQJC/0v+ihYHZAKuZsYAsjaKYHJ8GhO
seAY4odEHNgo5XA6RVOxnWnhLAcNp3meqAOP0ItvjHM/gQPzm0AUH7AtlUIB
iVlaNknmQ9mf8OqJ3KGw2jhZ2UuzLtxKRsNgf3wLOHDOiYl+g+HSOWlpojV0
cbKE5aRQg2cwVf9yx96wVneQ3aMDKsMyVPweaziA9otDaekcMYzmbEF1kRJY
MVbRFAfsAESnWCZS51xSqcUPLmIHR0WqvBHUUI5DWtc2IgJCllEFoyR52b/W
r+MsdeZTAVb0AiG3pgpx9EWsmqmzh24V34epO/HsLzjGRqqpB30sKJ2M3LpU
D8Mb6UV4CudPtjy/WOvFRdvpPXLcms/zpzYfsqTwWFyhQS72gJQn8FGTXTBY
qKo0tVm6GmrZ0lC001WEkpfSawVU1Z+1HGUY73TpJYhP38R1zqXKGX0Hl7vV
fKbVEoT9E5+p1Ru2vwgb0VTMa4MInlYCdlD6USofiKuZiCpa0ck9F/WIikK0
MKW/M2qKVWUuiRmFr6heFLTm8GXUQwNbonFy1p2v4pCvFxDYcuM9x/FIXnFH
PwGyae9Tj2xZtCy21IXtF1r+46jPAnbP0L47dL+gkoh0ZV1DHr3s+iErtSY4
lWtwIZFT13E1rMReJkwmuQLklNQjLs4N8ICuwE+futu2coMV0zk+Fzel8hBY
QdN7HnYUIhUNlagUls2q0weqD+H9U1KJByvEX4/TPUopdDmEoKpzha7xq8vR
AfxxFAXE5gMzAHHq4yID5oHSKAckSnorWaphzEw7J5BiURRUoAh2hwW2CQr9
4K6G1z1FJ+Cq4XUxWTikayGCa5w0J19MJB7HLouEIuolbGFXTCLm4Yr7UYw1
CFQYGORiorpuRot24IaEZIlx3+/qr6OTk+EFHcj1TP5OMTDTYq0GTb6w3FYl
efI2LbvCMlvhRVJ3SJdqILCzSYc+vuP/wE2SL5dq/XMutrzvKic3XDGFLaLI
rFBEghcJvLk6iuqAv0g+6S+BUhr+90t6j+GZ8SMkr+meGFp+SX7p83/6/+3/
Op63HsEYMPCIBhf/zfFxa9Jb9l9iHoI+Gh4NR/4NGoN/6P7vlzAtSB8BKhz+
pjF80Kx7NDoetccY8rscFdZHTMCsXrt7L0ejcJZ/bi+n4Zn9k3s5Oz85/aq9
zP/z72V4dnh+Ho9xRD+8eTt+j/s4OX3/dnT8/tXw5P1Px917OT4PZ/nn9nL8
/2Ivp1swxnvBjgJ0Ke+Hh++Bwrt1/MfsZXgYbOaf3cvxRRtvGcbE3nwyHMVj
dOzl/PAwbY3x2/dyevTb6MeVEwjdI4yd+E1jjDkBYMwxTOkOfBn+nu6Wmcn2
Xv5jcO43j/Fb9qKc8T/tXrbhtEWDPj1Pv4nESm7R/Mdn4y9IMM9+xVSasHKm
F6AleKJL9KD8GV8gcOLk6r6Y37+irDb+Y3wLyNIfHZ7zP9nGhUUoxPajS0FR
Nm6Y0M4ouY0FcMz+2ZVDQs3ipOzHVspQR0Sy/aqI0F5YozM2Wk+wNBkVSNPS
T7FiNODMBVB5nHkhtgUmnzEuRs6SQGmJ9DB1HIDoTNXt6dZQbpRMhmyDRivb
UZ4tMvPz8Ozj3gqxDULqMVHE2TabLxlHUVE3E7Lh1Zz+1yoJbeHEtIAcNx2A
F0TSdb53V0DYusLE6Lp3lg22Wn1uJUlkcf1u407VG72CnnLUP3qrliVfm/ys
5eGWy3WpJoOwCGFQjmhHNpdcMRpqpHs1XqFYyLGSdhPrns6qoyYnmUKTPyaV
T3EnJYslJ9alOVDGK1tkFtJeSrm2EXeNcagUgDg0fL9J8ZZYjSgOimRR3Ie7
nyQ4rzp76jZ735RB5su2n8XVAmYFLGxJzclCOEmuTW+4+bXq9y7rCFM4VhmZ
052p0QGNViEyiQJDEPrhTEtsz9TKchpFHVTWDzcnYcKYRorTJqRCTg3pjo26
2rgrizgbt1JG2+cQ2S+ktDa5dNjdGYeMBC3pNAxFrbNq/fRKviZbJXpCrdy0
7cEYUdgHd8M1jKt5V55V7iJVx2rrjqoGA7Wmqgt9+LavX3F5hJAYU1Reqxrf
TXUf1/XSYl6xe7Qnlj5ZmpskwRQw74+1nOUiLqq8dHTh8xUEg/7eVD6QBrAA
y4k2vZxHZmYHWD9JWc/cum56hEi8CrSyc+IbxSk574DaH8VfEpRykBB+tkH4
b11ZVQeUYZ+hriBi0tzJfkHd5RKKRhbjt+vhp90MrWvazW37mjClCi0emqDA
nmCrxxZ2aEhC9yYhnB4IYDq6TiifKt6g9A6qw2aCUXz0JiApSSsygvcSQEnU
kIvirnGylwD7fZAwl7kEZV5OKowcC+kMiDNvNuvDC2otCIQ40fLBeks8dJi7
71OgrGR2ZDX3gO4AUspTZEKR6Fm7lq3dcN1xNmTv7kJOMtEkXIKk2PiGz1gU
zIR8oYwo00ya323lfsLR3nVO44uF+GTyiMSQMY+DyTibUkCKqLLzqxHkuACS
iZZCFdEVcKuab7MTSn32cREg3sy4V5lbfuKDLf3NUbtQg1W81c0wBdUrozwq
IDJUL0QatMDNWsQVKu+IsMdIb428lVAd/LyiJn29UIZQsiBoSvnYWtNBYqsA
YanyCIHgvMbu6Ry+JtFrfFwC1Kt1vRJzd/d9Ax+Eg509B+iuLdWHQr76yAXO
TfZBfFp56TNkOESJuCPrWVxJhfs/wk9Yw7CmwgRyc5K+F6f8T/JMHYv+Min4
yuc3ouWxwh4dsKqoXkPXd9RyLvSI5NRjsyHqS6LvtJDuN2Qwn5msiCDJ5bq0
G8+QjMtSRAl0USru/mzqqo9up8LMHlyprw139+LoHDwj97FOPDbF/J4S4t6O
b8ZaYzZu4XA6GA2GHOnxAhnQIl8vuTA+Xqqn5u5ObRgwvUVjoo6m2HICtUBM
XsUiCKGk0rJqo9tOJadWV1QnPTjB+GNDLc2wRlCoDZIC6FpTDEY+Ys73cpBV
BOCkRSXgF4r5XOAZzxx6KAXZPdPR4HhfOL2P/Uziku4UmiUb6UuM9SQvg4hv
1aoyVYvd0ScenQJ2zF0gEMYRAa9RQ2ljtPhaqcCrUNfIlZp+Z+oScKXIgcTC
CWhpNfw4ERkTZPYpxc5XlLXHS6Ha4FwVjPoEU2C8zoq1hFZcJMRJJUghB8mP
JUcKo9pgfR3kJoIzJzVwxCbNb63j8hkAW8WVjyKM8o7jz0kiMPPwYCWwLSSD
U56oymOGVcRIGI56ANNiOXpXbhidEAL0oOPu3X33PbV+8n1xKVzM93/bCxvA
DffJ+SBCwF84wRktOcEhXvpD/AVD7XlHP7gdXeqOvAvicz6HL/zgHBEtB8Qv
6eFgNLo46jAZHQ+OToe7DEp+pNMTP9Lw4visY6STwfBo+/nWSOdnwZpOOub+
BQ734vBk50howtKTfO/OT8xYt18GYKn1/kUp94GqQ2kiBt/u4Jl42SXWOX9s
MRUfB3714uW+g3YXy+2iQD30E1pj4NbbNzDTe9Rx3sPHAG12Ad/2fAWfeC1a
Ed/vjMqyOwuPlHSxvicY76hLKWCZmSO/KRZ8SYoSrI2i1jX8yDxnHZFoFqjQ
czq4ougKQtL5dfqjJCB6e8eB5x2AizHpxqXkhBAcwmAIRTHGfNXfYY5hAGce
KUKwDgETvxhFXxwfsq8N/350Mjx2z49PT/WLo+iLk+ORm+P47MzNcXJ+fqRf
HEdfnB6fjdxbZxdujtMLRjD84iT64mx0PnRvnQ7d12dno5F+cRp/ISjGX/v5
zo/kOXxxFn1xfnTuvzg/dHs6Pzu60C/O4y/OTtwX56Mjt8KLw1O3qovoi4vD
Y/fW+cmZ+/ri6NB9MTyMvhid+i/OT9w+Lo7P3Q0Oh9EXx0eHfr6zU/f3U7lN
/GIUfXHi57gYnfgvzoZyg0iTYvR1BCnAT+rOQ1gRlrUIVPow0nQ3su6kCESd
bspH49tY4Eo4cY8kMLQNrpfddoHSa0VU9wTHzxJ0tddSIwIXzsFMuiWQbtP/
9se0KPeGaV9qNuynB/pgte9an5Bgy2H0W9nXpevwOFmDzOZS9Z5A3mz6FESu
FhQqjYgGlKiJlpLsTCzO+AlRFBaafys5+VpacnH4XxgUQiAOgS1AB3z9RF4P
oWt43Pkpvn6hr3v/bzoKPz31r7cAkKLFcqqsyGB44++545YF2Fa/BVY/D4XX
FGYjT3gaE+Rt4skJrMOmQ9CeSEwlaMkF5zeWD88Tf2WcAArHEEAuLgROq+dq
YnLtTubwIJZQMJUX4i5X1JLwY3qlGni4PTLpwULP4WZac2SJlN9vn1SGZQlg
3tH/xorjvbiSiVsJ5jg1ZP9IKCPVtqXLezLaDak2NefYYMu0rXy8RtqE4BCJ
qDNbAhGofoL4LJdF7FZEh9Bkl7hAUVJbbBMsl+dSrW6TG0w3zjj2joqnjg6D
o/qWh4Lz65ZQfMIOHKb1vdQbDgtES0FCB3k8GPV8/gzVMlJTKB+0EBw1gmt8
Z0Y+H7U2sD4e27HYgaHNFNgBFBcX45z/CeriIKShWQYFJLyqFO4q6M6wXs0o
l83d43tTw0FkmFEkufsUQZXQ1gmocM/YzuKJoV2jYCmci47UCYlk1CPii5n1
cJHHveSBu6Kcj4aetlL5ZSc8wRFKH1Ol91r6BB0PElqd0OqWxkgEW+tE0c1D
u/X1VntbdJvkTUIhiUUDzJEiP1r6iTe0S5/qsZsAEMsHJ/KBUguvRvKUb13w
+03YvhRBOqzuGAfSvmVvK+nvKEjLuewy5adiym8nRqE6b6Wwfmib10pHiejP
Tp7nfCMfxb9D3v4pbMklvfY+ArSy2QeTKtFci5lrPU8Pneck0Dioo6vRmHFa
DerBEZPOtgX0JMz46ujw2bZQCGsVmFuXcyxlxrGbuDJnMHbmIAwW91kLiAam
LsN426Au/Y6u5bIbCfF1x5NQrymvzIU7iRLFQs85wUG7dhf2B/+VUu6oZalf
kS+c3etcmO9Tqg4jb0SLc/alnFaXia3dM5RreUsfLSno54oxhj1I2SkWthTl
YhnsVXTpiJ/xqIU70GL5201uup1hcJQvwreIej5yhH5gStWKA0yNqdaS1UBq
ryXOKu5hKnUTO9bKTXgfOa2iRopG6BcjizbZGa8f0KnoyMMEhMYF7DOsqahO
xCAQRGI8ut52Us/uuO3uKA+u8JlEFT7j1ootb18ae/t2OnLujJRvTyRwQTrx
YRwJ5a9RzipnHmPuysSoEN0iMKG1cplbrXXDHIvKjCRst/doHJeufpNPau7y
HEXH6Mm129V4XxBJnlRxPLIZJh17DuhQWO+KuFtFUsqUqxG16JsQKgqBMNhL
PLhaYL8seGDOes3By67wmKgQTgrIKSCHq/S5iqLb9o4BZYIGm9U2ybVZcNnO
lBzjVlPCd8NXzwOiFeKGZ5VQqXlp8OEqoGk+JjMfNM8Pqco2QERQ4iTczfPk
fhHvT5PBHZHytYHFD7NlShKY8zPwgrpu0LsfZZEaVK5FCz5SLeFmveLShNRz
vCaSOKVSLc5VTYkKe1qmpV2otdeO4InKGnLqC1V/ILk6KJW730tXAPAJuyU4
dgclKpIIAoC6QWibY4g+r9wjVhTQFTvrna824erUkmkLagrKtb62QxDJ5EKH
6g5AAzgbgYLQAaXPE19c3d+LEpwtq3ssRWAX2VaSdEdMxa/7vSQSWSIoatfh
IE+/OtKyHbeNJ8+0MKzJk3BPaSk14v20HXbbUEdji3+HLjToLMopIrBU0e44
0hQ+z1Wmq7ifky9zqolRAQHbpgs9yY0L3TUCp5Tarqm3Pv6qF73LjoqSFG4A
FfJYYs9AVXUbDalgAk3G5FYNTc1NRc96YDTBzgwSTAM8YFFN1SmSl3B/dosK
US3vOPaD3uR+FnClKSXXcLyZYEZPj5URMDDoaLEHEvFm1XQt7RKpc0RtYqN9
bnj37NClPHMMuMkDIVCTieQJ0+uWwiJUyGl63F9qzrXr1TmpFRb9wSeRj09C
GNEdqF5fOecW0lltF9NBLVUAmBCdooNC6unAm7FW1T8hEaXJ0JXvyjN2Wb16
UqsGlXiJwEF2JSQT30azEnnMEkGXLrORxNXi+3RnalH7JvW9C0j3Cpq1z6U1
wY+c5xdqZiovfvrGy+f9uNf7rxQHhsHPd5XWkYSd3IG4/JQVW8FcQSHrktPa
WdsEWXMz5ZYukViWSDt2ZDzCbvNqJl1X6nDGmmfkdre5z7nGxE3tqUhNslAa
TaLeA9LwkISoOIr6QZoas9bXpVEwn6e6MgncufQio+ROyWf0PSB9K5+gH7Tk
snm6w+wYw7WtIgrVy6vRFIft7HGd0fZ9nHbckCOKPy2yjamTMCtWKisHXRNi
5QR1r8KldZkw2qEziHVJDWCxIgYqKdSQk6Q110RUKwgBRjwlrvtCT0qDcXwL
dach+wL3Hkmria3oL5wXK0Y01SRRiJXOXgm3mcu4lJTgHu/xb7+z2KMqQ/d3
q1HEHG1pmNQvREYt+tR1UyOpA21A60lIGAR3nFFmYFAbwvNBVZpyEUNZnYtj
4uUkrpc1stsg+9GZ9iTpt7VYdedpyE9oJqPYdi2IXqFEv5YGiaSiiYnBG8Cw
WY+lrua5FOFJKLV9yuCzI8G4x1evnIvsQHEBncd1gbest9OKjetCId9XKmgZ
77qQcvWQgvGEjEykZ2MoSXgyVYHMPUhc9pKqM2JyJxCiU9L0SMkJIQ9Mymav
WA/PXE0+rkPQw0F+ZGuXj+ibLiraDPdJlcwGZ0/GXiJE7dyNPaxBHkClgZZk
tA9DOz4c5pznhZNOBWUf88xbFsSSSVQ+6vX2Ut/4cfWAxXq2LW+evutov3Y3
XmELi4+XmqMt2RlvHFEO1iQzYgw5tjT1BLcFH72gLBKQhTYBDlvllDPJIOAy
5XBnrrlzxmTa9RbO59L13sMTgj8JQZKe7DqDYmO6nuwqOtLE0YLGmmIucjJm
eoSJHpLszHKRRAFikQgsbFCHcMPAKGZSp1OTYL7R5nfuKJYZwnMFMrbv10H0
dymN71GnbKLsnbABJSt13UVDQouTa8ObxQjfvsiQLOrOkyDW3VEhBygdiI5N
fagJIIUpZmVo9KPWYaAsTGojfSE+F3PqzGt3GA5PUfH7Ls407Aq6jSV4Qe5h
VOsBWBSdXZl09ziJYFGaFlq5OMfZ4Ew0eE1DEwVj1hRtSu3YnJE5uOYAi7+J
K0c4NL6MmhqS3cLF1JqgySrBuL4r99I+h25KHLX/ymbVqiG2sjPmV7LLZ8bb
GO+JPwf5C8Sd1b7iBLAgcVyanETVdODQQO/tu54LYYmFntTpZdWhcV1DVLZq
Y1jQG7ADyVNk0zkWflmLiMa5BZFKTWlrh1J6LOwciY2Q+ssmB7mbg9pRIAQ9
q8H6PU3Vd7xxJ10LphTJWPt5k3xgtULWq/H44M3bMSzmX2Ax5ycn6Hh0wpUv
eeTgHOWDm+v7l1SxCDHqAdSSFQusrmqotF/cyez14DQPx3cnIOs7A7W6iabO
ojZGDjtfF1RprD+RBn0t4PGj696Unoi5B7OLdb9HF0OKBERLfFXNfCS3jWsJ
bqH7PVmuffUF6qXsUmUiAkAmw7gxTavzvDJ5BvvuCiAiCoamJXJ2RHVFai30
S94VS2iTB7Q8b3XVCElsSxSn9Czlg+rrzbl6dp/eA/G6mg0SbBDzd5Srqye6
fQrA/txFufr4Qdljqr9jzArp85SaPfKE70rTv0fLi8/J3Ht3P95nMcsX+UQi
3ZOyx9xmjCQsUbThC9bj2eqlN5spboL2jDJe4vm8SOMZD7RN5fBXGVXI28Sb
iqSiknNeukI1iCPUPoTOJ5TAuTT+IHFNeo+OEPddvTaMnEByToqaecg1vZVb
rWPll0ZTHqSTfaiLULOuz90IZkcS14/7IHWltDpD2YTkRU0QbDukSCQW9Obu
sWhXR1bfjbtdHmHK9aW5FtkW4HnY3QnROUUb74LouI4kOmkrY8vfNcKoFtJb
FLQEkn2CjiXMFdpXSIYgIDk0MNEYHAhO1kFVd/JOuzZUJo4C7L5I9dapVFCr
WJMoArvLMlXb1xmUYUrkZn5npVm4mMlsmm27Fr0yQoAe956INOYuKuJsjX11
cXBfcWlZ0waEoBngjiCBgdRgZxqq5LDhbTuI4zuK/NjYMst7D7YPXuh00CGg
o1+WlBilkyBUxmwXbVKFQQqOM1eUw0rskFNZKISqQ3pFjrpEQywmx5BXTfI5
6BxXOXe9UvbIcQzpG5PV9JwK0F7XNRpP99723/x0jRWdKfyd24Vz2U0uSeiG
Fvt5LD1IwjFFCKCXlayu/tJaI7KEMEiCMAsjfQ/E2QWqJ9msbYOn1U6JgePo
5g50kbQTik2Z+QZlYScyzj9JJe2LwwWBj1ZSjns3FA/IeLglrpUzlXkmVRU5
/Ltd+cTEnIHMO+7FIWm44R4NhVX9bFhTEx8XbDZCymY+koP10Wi/ZpS/yN/v
q2P5fnmxzMpuYxUkQi+/iLgO5DisYvdmOA1fLed8zdorS9t4tUqS/SUcwsdb
FU8ZcEXu+bx1HLSRKKyA+H6mqc3uYvKlK5UGyDxdcAiQZOdTAh4HGbBXWNJN
QNPn72BDUulvJjEXBO/kS2j8JLAx7jQRnQYegfNvtux9mKmHLcSlXifbPqJu
bRyQypJy3NSOCJl2I9US066yxK+crOKzwX3FBReTQWAeq5drCq4vZWvSQVaX
GfoJQQF8tZnU+SwQpi6lCgIGRpHe54uGYbjYg/on2onkaH+JRAWndt3++eA+
XfA82woeFxhwrRVdL1F3IVmkjPaC0uFZnMlObo1u/iYuNsl0whZ1lIxpczT7
SnVS+keYdmupmZ/LIFbt7OyCWA5IufE4Wj5C0Zp7QcXhLnHLXW5pNHONlbEI
N+YXEseIWEfQ2ZQi9khDYBsv2l0/cl9Yn4xOZ1HAv4RzEQAoqIdEzZk6Y/WU
2yJ3HoorakjXYHzEQ1TkRNohkaztzsDtIlz8lhVSigAE7UDj6cnQGgZxuWaF
4rnJugscxOSZS76HlI3oZlRWT/N1pSQxWwLb7Rm5YKeVlniN90R3Q2Kr/yqy
QFMvtRwnCXG9llDXU8PHz+K8a1EmKpHQ9l6LqTtMhY4+M9wx2jqLDnkIBEtj
lHNZqEGmKtbgiF/ScxDoyjgwmwRIjenGkAlfTddfTPwpmZdvLn+4bLkPRc5S
T7DHiLLi17kmOZunfXXBrTHCCojb/smWHyTvDjtUT+YcFqPZq4kaSAJvG4D5
rJCKFq0qjex98m1d1ajIkbhhlx5KWcaYdnE/4X007AX2tXuxCZbbl0QsEAEL
B3KunoRmkcgf5xTx1opA9pGAHoX63bWiEy3SPuYPzUfyqoQKqDOUiYfSSG47
NuNgX5HP3neVOHMM52St2zs4OS7bJwPwZERLw5MCeRWXKPTWnxs5odYlgnDF
fXRZzgLYQQezV8bQcA0KjaVUPtURW3nDStSiVHnRojo7JVOf0UtM0AWBDyhE
YzqzTVPONkXGyBVYXMFTbpknJ0Rxx76lE1Af9ggyKWQ6h7EXXMtXPL15q60Y
+fEXVSEOY5AZsUlDN03BkHD6AJYXNx4N1LIv1OtiQSeo2cXSyBitRVcLbNNb
pLd8XWSBxh/6U/lB3Spw6UWF2tZiwzVCsbbAV/Q1d0ZajocMyhAAdaYavEHt
Fay4D6sAgMkeSuoNbpa5lb7frrucW4KvJoIRmxKU8pEaJmC5IptzRigGxFC9
nRA9U0WKQfo9pduwdsQfBTY513EovHepVPahBb7cRd75kCiXQkyyNjzTlTvs
XV3txcelIQq260akejUCKtfUcu7ppgLOS8x7QyifiY1Qw76xFg7TDYfkPB/b
NBGfUZOYsEeuqKoPmHJx1eV34rdBHqhqIApUgwOuj+JW11wNRKmjREhFW2l5
DaU21SO3M1JfDEm/DxLrIZF6IvZIvQG64S9AIsD8dgEt544m+TEOad4mPoP0
XelaIKtFuPNq+ND1uFn8LPVIGy2ggcEUD1xApEXlPJtgy2q70XVQI6HVTYR3
Jz9fxM1GThjz08upFr0gSoJ8Ois/0On/q8nK/m1uQHhNX+bUgPQun5Lt4rVp
msKghNtL3yI2vQPVhzrxwTuwsvQVdj4zILi8NuYBoPJFZvPiMUu/y+pJZbNe
+goPyab3drqo5qbMQXa+BFyp8RUAg78bg/aVRYZ6/HfVeprNsrzG4RoY7T4r
sjVgccYNHW+qErjR23yRFVOTufCoGdbuKii76zE3T4Pk/wKmLLPDdN4AAA==

-->

</rfc>

