The exponential growth of wireless devices and stringent reliability requirements of emerging applications demand fundamental improvements in distributed channel access mechanisms for unlicensed bands. Current Wi-Fi systems, which rely on binary exponential backoff (BEB), suffer from suboptimal collision resolution in dense deployments and persistent fairness challenges due to inherent randomness. This paper introduces a multiagent reinforcement learning framework that integrates artificial intelligence (AI) optimization with legacy device coexistence. We first develop a dynamic backoff selection mechanism that adapts to real-time channel conditions through access deferral events while maintaining full compatibility with conventional CSMA/CA operations. Second, we introduce a fairness quantification metric aligned with enhanced distributed channel access (EDCA) principles to ensure equitable medium access opportunities. Finally, we propose a centralized training decentralized execution (CTDE) architecture incorporating neighborhood activity patterns as observational inputs, optimized via constrained multi-agent proximal policy optimization (MAPPO) to jointly minimize collisions and guarantee fairness. Experimental results demonstrate that our solution significantly reduces collision probability compared to conventional BEB while preserving backward compatibility with commercial Wi-Fi devices. The proposed fairness metric effectively eliminates starvation risks in heterogeneous scenarios.

IEEE 802.11be (Wi-Fi 7) introduces Multi-Link Operation (MLO) as a While MLO offers significant parallelism and capacity, realizing its full potential in guaranteeing strict delay bounds and optimizing Quality of Service (QoS) for diverse, heterogeneous traffic streams in complex multi-link scenarios remain a significant challenge. This is largely due to the limitations of static Enhanced Distributed Channel Access (EDCA) parameters and the complexity inherent in cross-link traffic management. To address this, this paper investigates the correlation between overall MLO QoS indicators and the configuration of EDCA parameters and Acess Catagory (AC) traffic allocation among links. Based on this analysis, we formulate a constrained optimization problem aiming to minimize the sum of overall packet loss rates for all access categories while satisfying their respective overall delay violation probability constraints. A Genetic Algorithm (GA)-based MLO EDCA QoS optimization algorithm is designed to efficiently search the complex configuration space of AC assignments and EDCA parameters. Experimental results demonstrate that the proposed approach’s efficacy in generating adaptive MLO configuration strategies that align with diverse service requirements. The proposed solution significantly improves delay distribution characteristics, and enhance QoS robustness and resource utilization efficiency in high-load MLO environments.

With the growing demand for next-generation wireless networks and the rapid development of AI, traditional Wi-Fi MAC design challenges to manage the increasingly complex network optimization, and the machine learning-integrated MAC (MLMAC) initiative we work on is expected to contribute to the future intelligent network. This paper studies deep reinforcement learning (DRL)-empowered Wi-Fi distributed channel access (DCA) strategy, introducing the concept of network access entropy to effectively quantify DCA chaotic degree and facilitate ML-MAC performance analysis. We focus on the centralized training decentralized execution (CTDE) paradigm, e.g. the algorithm QMIX, in the multiagent reinforcement learning (MARL) framework and formulate decentralized-partially observable Markov decision process (DECPOMDP) in DCA cooperation network. We consider the tradeoff between total and individual rewards and first introduce wait decision counter (WDC) as part of the DCA agent observation, enabling deep neural network (DNN) to achieve better performance and convergence stability. Extensive simulation results on our MLMAC protocol stack platform with NS3 and PyTorch demonstrate the strategy's superiority over CSMA/CA under both unsaturated and saturated traffic, as well as the advantage revelation of i) trained-network generalization; ii) dynamic access robustness; iii) protocol heterogeneity fairness.

This paper presents PipeFusion, an innovative parallel methodology to tackle the high latency issues associated with generating high-resolution images using diffusion transformers (DiTs) models. PipeFusion partitions images into patches and the model layers across multiple GPUs. It employs a patch-level pipeline parallel strategy to orchestrate communication and computation efficiently. By capitalizing on the high similarity between inputs from successive diffusion steps, PipeFusion reuses one-step stale feature maps to provide context for the current pipeline step. This approach notably reduces communication costs compared to existing DiTs inference parallelism, including tensor parallel, sequence parallel and DistriFusion. PipeFusion enhances memory efficiency through parameter distribution across devices, ideal for large DiTs like Flux.1. Experimental results demonstrate that PipeFusion achieves state-of-the-art performance on 8$\times$L40 PCIe GPUs for Pixart, Stable-Diffusion 3, and Flux.1 models. Our Source code is available at \url{https://github.com/xdit-project/xDiT}.

In this paper, the problem of efficiently evaluating the error performance of linear block codes over the AWGN channel is considered. Based on the geometric structure of the channel, we define the l2-norm of the noise vector as a random variable and refer to its sample space as the radius domain. A minimum-variance importance sampling (IS) estimator is proposed by deriving the optimal IS distribution on the radius domain. The IS gain of the proposed estimator compared to the Monte Carlo method is analyzed. The asymptotic IS gain for high SNR, which only depends on the minimum distance of the code, is derived. Finally, the effectiveness of the proposed estimator and the accuracy of the asymptotic IS gain are verified through simulation.

In this paper, the problem of efficient performance evaluation of linear block codes over the AWGN channel with M-PSK modulation is considered. Based on the geometric structure of the channel, we define the tangent of the half-angle of the circular cone, whose central line passes through the transmitted signal vector and the origin, as a random variable and refer to its sample space as the angular domain. We propose a minimum-variance importance sampling (IS) estimator by deriving the optimal IS distribution in the angular domain. Besides, we derive the asymptotic IS gain of the proposed estimator compared to the Monte Carlo method as SNR tends to infinity. The effectiveness of the proposed estimator and the accuracy of the asymptotic IS gain are verified through simulation.

In this paper, we propose an Importance Sampling (IS) scheme for fast simulation of linear block codes over binary symmetric channels (BSCs). By re-formulating the IS problem into the one-dimensional Hamming weight space, we propose a novel IS estimator and derive the optimal IS distribution which will minimize the variance of the estimator. Consequently, a corresponding iterative algorithm is proposed. The effectiveness of the proposed IS algorithm compared to the state-of-the-art IS algorithm is demonstrated in the word error rate simulation of both LDPC and Polar codes.