Integrated sensing and communication (ISAC) technique has been viewed as a promising component in future network. A major challenge for ISAC systems is that the constraint introduced by sensing functionality will constrain the degrees of freedom in waveform design and results in large multi-user interference (MUI), thus degrading the communication performance. In this paper, we study the employment of RIS in mitigating MUI in ISAC systems. For practical consideration, we investigate joint constant-modulus waveform and discrete RIS phase shift design, with the aim of minimizing MUI under the Cramer-Rao bound (CRB) constraint for direction of arrival (DOA) estimation. An alternating optimization algorithm is proposed to solve the formulated problem, and two schemes are proposed to deal with the discrete RIS phase shifts. Simulation results show that the proposed algorithm can dramatically improve the sum rate, and the performance under moderate phase shift quantization level is close to that under continuous phase shifts.

Non-orthogonal multiple access (NOMA) will provide massive connectivity for future Internet of Things. However, the intrinsic non-orthogonality in NOMA makes it non-trivial to approach the performance limit with only conventional communication-theoretic tools. In this paper, we resort to deep multi-task learning for end-to-end optimization of NOMA, by regarding the overlapped transmissions as multiple distinctive but correlated learning tasks. First of all, we establish a unified multi-task deep neural network (DNN) framework for NOMA, namely DeepNOMA, which consists of a channel module, a multiple access signature mapping module, namely DeepMAS, and a multi-user detection module, namely DeepMUD. DeepMAS and DeepMUD are automatically trained in a data-driven fashion, and a multi-task balancing technique is then proposed to guarantee fairness among tasks as well as to avoid local optima. To further exploit the benefits of communication-domain expertise, we introduce constellation shape prior and inter-task interference cancellation structure into DeepMAS and DeepMUD, respectively. These sophisticated designs help to reduce the implementation complexity without sacrificing DNN’s universal function approximation property, which makes DeepNOMA a universal transceiver optimization approach. Detailed experiments and link-level simulations show that higher transmission accuracy and lower computational complexity can be simultaneously achieved by DeepNOMA under various channel models, compared with state-of-the-art.

Tactile Internet of Things (IoT) requires ultra-responsive and ultra-reliable connections for massive IoT devices. As a promising enabler of Tactile IoT, grant-free non-orthogonal multiple access (NOMA) exploits the joint benefit of grant-free access and non-orthogonal transmissions to achieve low latency massive access. However, it suffers from the reduced reliability caused by random interference. Hence, we formulate a variational optimization problem to improve the reliability of grant-free NOMA. Due to the intractability of this problem, we resort to deep learning by parameterizing the intractable variational function with a specially designed deep neural network, which incorporates random user activation and symbol spreading. The network is trained according to a novel multi-loss function where a confidence penalty based on the user activation probability is considered. The spreading signatures are automatically generated while training, which matches the highly automatic applications in Tactile IoT. The significant reliability gain of our scheme is validated by simulations.

Non-orthogonal multiple access (NOMA) promises to fulfill the fast-growing connectivities in future Internet of Things (IoT) using abundant multiple-access signatures. While explicitly notifying the utilized NOMA signatures causes large signaling cost, blind signature classification naturally becomes a low-cost option. To accomplish signature classification for NOMA, we study both likelihood- and feature-based methods. A likelihood-based method is firstly proposed and showed to be optimal in the asymptotic limit of the observations, despite high computational complexity. While feature-based classification methods promise low complexity, efficient features are non-trivial to be manually designed. To this end, we resort to artificial intelligence (AI) for deep learning-based automatic feature extraction. Specifically, our proposed deep neural network for signature classification, namely DeepClassifier, establishes on the insights gained from the likelihood-based method, which contains two stages to respectively deal with a single observation and aggregate the classification results of an observation sequence. The first stage utilizes an iterative structure where each layer employs a memory-extended network to explicitly exploit the knowledge of signature pool. The second stage incorporates the straight-through channels within a deep recurrent structure to avoid information loss of previous observations. Experiments show that DeepClassifier approaches the optimal likelihood-based method with a reduction of 90% complexity.

The deployment of mobile edge computing services in LEO satellite networks achieves seamless coverage of computing services. However, the time-varying wireless channel conditions between satellite-terrestrial channels and the random arrival characteristics of ground users’ tasks bring new challenges for managing the LEO satellite’s communication and computing resources. Facing these challenges, a stochastic computation offloading problem of joint optimizing communication and computing resources allocation and computation offloading decisions is formulated for minimizing the long-term average total power cost of the ground users and the LEO satellite, with the constraint of long-term task queue stability. However, the computing resource allocation and the computation offloading decisions are coupled within different slots, thus making it challenging to address this problem. To this end, we first employ Lyapunov optimization to decouple the long-term stochastic computation offloading problem into the deterministic subproblem in each slot. Then, an online algorithm combining deep reinforcement learning and conventional optimization algorithms is proposed to solve these subproblems. Simulation results show that the proposed algorithm can achieve the superior performance while ensuring the stability of all task queues in LEO satellite networks.

In this letter, we exploit constellation rotation to enhance the performance of uplink non-orthogonal multiple access (NOMA) with a successive interference cancellation receiver. We characterize the received signal with a Gaussian mixture model (GMM), where optimal rotation angle is obtained by maximizing the entropy of GMM. Then, we efficiently solve the entropy maximization problem by deriving an optimal closed-form approximate problem via variational approximation. Performance analysis and simulation results show that the proposed scheme achieves larger capacity and lower bit error rate compared with conventional NOMA. The performance gain is more significant with fewer receiving antennas or with smaller channel gain difference between multiplexed users.