Mahyar Shirvanimoghaddam

Short packet transmission is crucial for ultra-reliable and low-latency communications (URLLC), yet designing efficient channel codes for such scenarios remains challenging due to the limited blocklength and strict reliability requirements. This paper proposes a novel end-to-end autoencoder framework for short analog fountain codes (AFC) that jointly optimizes encoder and decoder components. The framework combines a learnable AFC encoder featuring attention-enhanced weight selection with a graph neural network (GNN)-based decoder, maintaining the inherent rateless property while enabling systematic optimization. Numerical results demonstrate that our framework produces the optimized generator matrix and consistently outperforms conventional AFC schemes in terms of block error rate (BLER), approaching near-optimal error rates and realized rates at high signal-to-noise ratios (SNRs).

In this letter, we propose a novel decoding scheme for primitive rateless (PR) codes to improve belief propagation (BP) performance in the short blocklength regime. We first design the PR codes using primitive polynomials whose supports form a Golomb ruler, enabling the construction of a 4-cycle-free parity-check matrix (PCM) without the need for auxiliary variable nodes. We introduce a second PCM derived by sorting channel log-likelihood ratios and applying Gaussian elimination. The double-bases decoder operates by running BP decoding in parallel on both PCMs, and selecting the candidate codeword with the shorter Euclidean distance to the received signal. Simulation results show that the proposed double-bases BP approach can outperform the multiple-bases BP decoder while reducing the number of required PCMs and complexity. The PR code paired with the proposed decoder offers competitive performance compared to 5G-polar, RM, and LDPC codes in the short blocklength regime.

This paper presents the analysis and a proof-of-concept design of the low-complexity early termination hybrid automatic repeat request (ET-HARQ) for ultra-reliable low-latency communication (URLLC). In ET-HARQ, unfit packets are flagged for retransmission prior to decoding, a process that significantly diminishes decoding complexity and facilitates swift HARQ reporting. This characteristic makes ET-HARQ well-suited for URLLC applications. We analyze the impact of ET-HARQ on the packet error rate (PER), throughput, and complexity performance in the finite block length regime, taking into consideration cyclic redundancy check (CRC) limitations. Numerical results indicate that ET-HARQ significantly reduces the decoding complexity and improves throughput with little to no loss in the PER. In addition, ET-HARQ demonstrates resilience even with a short CRC, whereas the imperfections of a short CRC significantly impact PER reliability in standard HARQ. To validate our analysis, we also design a practical early termination mechanism involving belief propagation and neural network (BP-NN) to predict the decodability of the received packet. Testing with BCH and CRC-polar codes shows that it can reach up to a 70 % ∼ 80 % prediction accuracy with packet lengths less than 128 bits encoded with high-density linear block codes. Simulation shows that the BP-NN-based ET-HARQ has significantly lower complexity at this accuracy level than the standard HARQ with similar reliability.

Sparse vector coding (SVC) is a promising coding technique to achieve high transmission reliability and low latency for short packet communications. However, for SVC with conventional combination-based sparse mapping, a small increase of transmitted bits may lead to excessively long sparse vectors, resulting in unsatisfactory transmission performance when coding efficiency is high. In this paper, we propose a generalized index redefinition (IR)-based SVC (GIR-SVC) to significantly enhance the efficiency of SVC. The IR mechanism enables multiple index bit streams to share position resources in SVC, with the help of constellation labels. GIR-SVC constructs the sparse vector using a hybrid IR mechanism that integrates the unlabeled IR and the pairwise-grouping-based labeled IR, which allows efficient mapping and de-mapping of index bits without requiring index tables. Consequently, the proposed GIR-SVC can be efficiently decoded without the index table using sparse recovery algorithms. Theoretical analysis is conducted to validate the block error rate (BLER) performance of GIR-SVC. Simulations show that GIR-SVC can significantly reduce the decoding delay compared to existing approaches, while maintaining the high transmission reliability.

Short analog fountain code (AFC) is a promising channel code for short packet communications, particularly in ultra-reliable and low latency communications (URLLC) scenarios. This letter proposes a short partitioned transmission strategy based on AFC (PT-AFC) to further enhance the error rate and latency performance of AFC at short block lengths. The key idea is to partition the information block into multiple short blocks, each being encoded using an AFC, and then concatenate the AFC coded symbols. The partitioning of the message allows for parallel decoding of various partitions at the receiver, which significantly reduces the overall latency. Additionally, this letter provides an analytical framework to minimize the number of short cycles in the bipartite graph of the respective code, where we determine an upper bound on the number of partitions, given the information block length and code rate. Simulation results show that PT-AFC outperforms existing AFC schemes with significantly reduced latency.

We propose real-time remote estimation of a dual-process status update system for mission-critical applications using nonorthogonal multiaccess (NOMA) and orthogonal multiaccess (OMA) techniques. We consider the finite block length regime so that, with OMA, the status updates of each process are transmitted using time slot sharing. Meanwhile, with NOMA, we use a power domain packet combining to allow simultaneous status updates of each process in each time slot. To compensate for the reliability loss due to short packet lengths and multiaccess, we utilize packet retransmission with hybrid automatic repeat request (HARQ). Specifically, we propose OMA-HARQ and NOMA-HARQ transmission control policies, where multiaccess resource sharing is jointly designed with HARQ. We propose dynamic and static scheduling policies by optimizing time-sharing or power-sharing ratios between sensors over time. The dynamic policy utilizes higher flexibility to optimize the reliability under restricted Age-of-Information (AoI), leading to the best estimation mean-squared-error (MSE) performance. We formulate and solve policy optimization problems, where both long-term average MSE and its variance minimization are the targets. We obtain optimal policies to minimize the composite objective function of costs of each process using the Markov decision process (MDP) framework and relative value iteration algorithm. An intensive simulation study shows significant performance improvement over existing approaches.

We propose a joint channel estimation and signal detection approach for the uplink non-orthogonal multiple access (NOMA) using unsupervised machine learning. We apply a Gaussian mixture model (GMM) to cluster the received signals, and accordingly optimize the decision regions to enhance the symbol error rate (SER) performance. We show that, when the received powers of the users are sufficiently different, the proposed clustering-based approach achieves an SER performance on a par with that of the conventional maximum-likelihood detector (MLD) with full channel state information (CSI). We study the tradeoff between the accuracy of the proposed approach and the blocklength, as the accuracy of the utilized clustering algorithm depends on the number of symbols available at the receiver. We provide a comprehensive performance analysis of the proposed approach and derive a theoretical bound on its SER performance. Our simulation results corroborate the effectiveness of the proposed approach and verify that the calculated theoretical bound can predict the SER performance of the proposed approach well. We further explore the application of the proposed approach to a practical grant-free NOMA scenario, and show that its performance is very close to that of the optimal MLD with full CSI, which usually requires long pilot sequences.

In this paper, we consider a federated learning problem over wireless channel that takes into account the coding rate and packet transmission errors. Communication channels are modeled as packet erasure channels (PEC), where the probability of erasure is determined by block length, code rate, and signal-to-noise ratio (SNR). In spite of fluctuations in instantaneous loss of federated learning (FL), we prove that the expectation of loss converges even in the presence of packet erasure. To mitigate the impact of packet erasure on FL performance, we suggest a paradigm in which the central node (CN) makes use of memory. In particular, we propose two schemes in which, in the event of packet erasure, the CN retains either the most recent local updates or the most recent global parameters. We investigate the impact of coding rate, SNR, and the CN memory on the convergence of FL. For both short-and long-packet communications, we examine a realistic scenario of a massive IoT under the assumption of error-prone transmissions. Our simulation results demonstrate that even a single memory unit has a considerable effect on the FL’s efficiency in erroneous communication.

In this paper, we propose a dynamic re-transmission strategy for the two-user uplink non-orthogonal multiple access scenario (NOMA) with hybrid automatic repeat request (HARQ) and more than one allowed re-transmissions. In the proposed scheme, whenever both users simultaneously receive negative acknowledgment (NACK) from the base station (BS), the high-power user is asked to re-transmit its previous packet, while the low-power user is requested to send a new packet in the next time slot. The low-power user is allowed to retransmit its previous packets when the other user stops the re-transmission. We propose a Markov model to analyze the dynamics of the proposed scheme and the conventional scheme, where the users always send the re-transmission when receiving a NACK from the BS. We find the optimal power splitting ratios to minimize the maximum packet error rate (PER) amongst users. Simulation results show that the proposed dynamic approach significantly improves the PER of the low-power user at the expense of a negligible increase in the PER of the high-power user.