SK(2) Coding Scheme

• SK(2) Coding Scheme is a collection of second-order constructions that use two-component structures to generalize classical protocols in feedback, fading, and cryptographic applications.
• It employs rigorous methodologies like MMSE estimation, recursive encoding, modulo operations, and iterative decoding to achieve doubly-exponential error decay and near-capacity performance.
• Its practical implementations lead to improved spectral efficiency, reduced complexity, and robust security in diverse channels and unsourced multiple access scenarios.

The term "SK(2) Coding Scheme" encompasses multiple distinct constructions in contemporary information theory and coding literature, each denoted "SK(2)" for leveraging a two-step, order-2, or two-component structure generalizing foundational schemes (primarily Schalkwijk–Kailath, Stern, and sparse Kronecker-product methods). This article provides a comprehensive synthesis of all significant "SK(2)" schemes in use, focusing on their mathematical principles, construction methodology, achievable rates, complexity, and security.

1. SK(2) in Feedback Gaussian Channels: Generalized Schalkwijk–Kailath Scheme

For communication over discrete-time real-valued channels with feedback and AR($p$) Gaussian noise, the SK(2) coding scheme is a second-order linear feedback construction generalizing the classical Schalkwijk–Kailath (SK) protocol. The AR($p$) channel is modeled as:

$Y_i = X_i(M,Y_1^{i-1}) + Z_i,$

with $\{Z_i\}$ a zero-mean stationary AR($p$) Gaussian process. The SK(2) scheme operates as follows (Su et al., 14 Jan 2026):

• Codeword Construction: Message $M$ is mapped to two independent standard Gaussian scalars $(U_1, U_2)$, to be transmitted using blocklength $n$.
• Encoding:
  • $X_1 = U_1$, $X_2 = U_2$
  • For $i\ge3$, define $V_{i+1} = a V_i + b V_{i-1}$, with $V_1=U_1$, $V_2=U_2$.
  • Transmit $X_i = V_i - \mathbb{E}[V_i|Y_1^{i-1}]$.
  • Parameters $a$, $b$ (or, equivalently, poles $\gamma_1$, $\gamma_2$) are chosen to satisfy the average-power constraint.
• Decoding: The receiver performs posterior MMSE estimation of $(U_1,U_2)$ based on $Y_1^n$.

The achievable rate is given by:

$$\bar I_{\mathrm{SK2}}(P) = \max_{|\gamma_1| > 1, |\gamma_2| > 1}\; 2\log(|\gamma_1|\wedge|\gamma_2|),$$

subject to an explicit power constraint dependent on the AR($p$) spectrum (via $L_Z(\gamma_i^{-1})$). The SK(2) scheme strictly outperforms single-pole schemes (SK(1)) on AR(2) noise (contradicting Butman's conjecture of single-pole optimality), and reduces to the classical SK construction for AWGN and AR(1) channels (Su et al., 14 Jan 2026).

2. SK(2) for Fading Channels and Quantized Feedback

The SK(2) principle has been adapted to quasi-static fading channels with imperfect channel-state information at the transmitter (I-CSIT) and quantized feedback. Here, the challenge lies in mitigating the residual error from imperfect CSI and preventing error propagation from feedback quantization (Yang et al., 2 Jul 2025):

• System Model: Forward channel $Y_i = h X_i + W_i$, feedback channel $\widetilde{Y}_i = \widetilde{X}_i + Z_i$ with scalar-lattice quantization noise; blocklength $N$, average-power $P$.
• Encoder–Decoder Structure:
  • Initialization: $X_1 = \sqrt{P}\, \Theta$ with $\Theta$ a PAM-mapped message.
  • Feedback implements a modulo-lattice operation with public dithering to ensure power constraints and bound aliasing probability.
  • Repeat: Forward transmission encodes the quantized feedback error, with an auxiliary subtraction at the receiver to cancel quantization noise.
• Parameter Design: Feedback and error-recursion parameters ($\alpha$, $\gamma_i$, step size $d$, etc.) are calculated to guarantee doubly-exponential decay of the estimation error and strict power/admissible error bounds.
• Performance: For perfect CSI ($D\to 0$) and noiseless feedback, SK(2) coincides with the classical SK approach. Under realistic I-CSIT and quantized feedback, SK(2) achieves capacity-approaching rates with error probability decaying as $O(2^{-2N})$ (Yang et al., 2 Jul 2025).

A related construction extends SK(2) to two-path quasi-static fading, treating the delayed (second) path as an amplify-and-forward relay. After an initialization ($X_2=0$ "primes" the relay), each round's error symbol is propagated and combined optimally via MMSE updates, leading to effective gain $H = \sqrt{h_1^2 + h_2^2}$ and capacity:

$R < \frac{1}{2}\log\bigl(1+H^2\,P/\sigma^2\bigr)$

with error probability vanishing doubly-exponentially in $N$ (Yang et al., 10 Jan 2026).

3. SK(2) as Sparse Kronecker-Product Code for Unsourced Multiple Access

In unsourced multiple-access (UMA) over AWGN, "SK(2)" (here, Sparse Kronecker-product, Editor's term) denotes a coding construction where each user's message is split and encoded via a Kronecker product of a sparse codeword and a short FEC codeword (Han et al., 2021):

• Code Construction:
  • $m \in \{0,1\}^B \to [m_1; m_2]$: $m_1$ encodes via a fixed-weight, index modulation code $C_1 \subset \{0,1\}^{n_1}$ (with sparsity $g$), $m_2$ by a standard FEC code $C_2 \subset \{0,1\}^{n_2}$.
  • User's codeword: $x = u \otimes v\in \{0,1\}^{n_1 n_2}$.
• Multiple Access Model: Active users' codewords superimpose linearly with AWGN.
• Iterative Decoding:
  • BiG-AMP (bilinear GAMP) for joint estimation,
  • Soft-in/soft-out decoders for sparse/IM and FEC components,
  • CRC-aided interference cancellation to hard-decoded users for error-floor reduction.
• Performance: Achieves per-user error probability (PUPE) within 0.1 dB of the random coding bound at code length $n=30,000$ for up to 75 active users. Complexity per iteration is dominated by the $O(n)$ bilinear factorization (Han et al., 2021).

4. SK(2) as a Five-Pass Code-Based Identification Protocol

In code-based cryptography, "SK(2)" refers to a five-pass, $q$-ary, quasi-cyclic identification scheme generalizing Stern's binary three-pass protocol (Cayrel et al., 2010). This zero-knowledge protocol is based on the $q$-ary syndrome decoding (qSD) problem:

Key Steps:

• Key Generation: Choose $H \in \mathbb{F}_q^{r\times n}$ (e.g., double-circulant to minimize storage), weight $w$, secret vector $s \in \mathbb{F}_q^n$ of $wt(s) = w$, and syndrome $y = H s^T$.
• Protocol Passes:

1. Prover picks random $u$, permutation $\Sigma$, scaling vector $\gamma$; computes commitments $c_1$ and $c_2$ (using a hash $\mathsf{h}$ modeled as a random oracle).
2. Verifier sends scalar challenge $\alpha\in\mathbb{F}_q$.
3. Prover sends $B$, a twisted version of $u+\alpha s$.
4. Verifier sends bit challenge $b\in\{0,1\}$.
5. Prover reveals either $(\Sigma,\gamma)$ or a twisted secret, and verifier checks consistency.

• Security and Efficiency:
  • Cheat probability per round drops to $\approx 1/2$, reducing the number of rounds for a target soundness error.
  • Quasi-cyclic forms yield dramatic public-key size reductions: e.g., for 128-bit security, SK(2) requires 2.5 Kbits versus Stern’s 2 Mbits.
  • Communication and computation are also substantially improved due to field operations in $\mathbb{F}_{256}$ and structural key compression (Cayrel et al., 2010).

5. Error Probability and Efficiency in Two-Round (SK(2)) Feedback Codings

Although the name "SK(2)" is not canonical in all references, the two-round instance of Schalkwijk–Kailath, and its modulo-variant, exhibit distinctive properties:

• For AWGN with noisy feedback:
  • Classical SK(2)'s estimation MSE is analytically given by $1/\text{SNR}^2 + 1/(\text{SNR}\cdot S)$ for feedback SNR $S$.
  • Modulo-SK(2) further enforces a bounded feedback dynamic range via modulo reduction, maintaining numerical stability.
  • Achievable error probabilities decay doubly-exponentially in rounds for both the noiseless and practical modulo-constructions (Ben-Yishai et al., 2020).

6. Comparisons, Extensions, and Research Implications

A concise comparative summary across key domains:

SK(2) context	Main Feature	Key Advantage
Feedback Gaussian AR	2-pole driver, closed-form rate	Strictly better than SK(1) for AR(2), disproves single-pole optimality (Su et al., 14 Jan 2026)
Quasi-static fading	Lattice-modulo, error-canceling	Preserves doubly-exponential error decay under I-CSIT and quantized feedback (Yang et al., 2 Jul 2025)
Unsourced MAC	Sparse Kronecker coding	Near-random-coding-bound PUPE, low complexity (Han et al., 2021)
Cryptographic ZK	5-pass protocol, q-ary codes	Smaller keys/comm, ½ cheat prob, structural compression (Cayrel et al., 2010)
AWGN feedback, $N=2$	Explicit error/complexity formulas	3 dB SNR gain, numerical stability with modulo (Ben-Yishai et al., 2020)

A plausible implication is that higher-order (SK($p$)) feedback schemes will further close the gap to feedback capacity for general AR($p$) Gaussian or multi-path fading channels, suggesting a direction towards feedback-optimal coding schemes matching the spectral poles of the noise/channel. In cryptographic applications, further reductions in key and communication sizes may depend on new algebraic code structures or improved NP-hardness reductions.

7. Outlook and Open Problems

Extensions of SK(2) have been presented for higher-order AR($p$) channels via $k$th-order recursions ($k=p$ conjectured optimal), ARMA($p,q$) scenarios by matching zero-pole pairs, and fading models via DFT-domain transformations (Su et al., 14 Jan 2026, Yang et al., 10 Jan 2026). In post-quantum cryptography, the main open issue is the security of structured matrices (e.g., quasi-cyclic) against emergent attacks not reducible to information-set decoding. For feedback schemes, the finite-blocklength regime, robustness to imperfect CSI, and low-complexity decoder design remain active research topics.

In summary, the SK(2) designates several significant advancements in coding and cryptography, unified by their second-order, two-component, or two-stage recursive structure, each exhibiting improved rates, security, or complexity and directly impacting both theoretical and practical aspects of modern communication and cryptographic systems.