Search

The continuously growing size of deep neural networks (DNNs) has sparked a surge in research on model compression techniques. Among these techniques, multi-resolution model compression has emerged as a promising approach which can generate multiple DNN models with shared weights and different computational complexity (resolution) through a single training. However, in most existing multi-resolution compression methods, the model structures for different resolutions are either predefined or uniformly controlled. This can lead to performance degradation as they fail to implement systematic compression to achieve the optimal model for each resolution. In this paper, we propose to perform multi-resolution compression from a Bayesian perspective. We design a resolution-aware likelihood and a two-layer prior for the channel masks, which allow joint optimization of the shared weights and the model structure of each resolution. To solve the resulted Bayesian inference problem, we develop a low complexity partial update block variational Bayesian inference (PUB-VBI) algorithm. Furthermore, we extend our proposed method into the arbitrary resolution case by proposing an auxiliary neural network (NN) to learn the mapping from the input resolution to the corresponding channel masks. Simulation results show that our proposed method can outperform the baselines on various NN models and datasets.

Channel estimation is crucial to modern wireless systems and becomes increasingly challenging when the ultra-sized antenna is configured in sub-6GHz wireless communication systems. In an ultra-massive multiple-input multiple-output (U-MIMO) orthogonal frequency division multiplex (OFDM) system, the channel demonstrates spatial non-stationarity. Additionally, the limited pilot location in the OFDM system further complicates the channel estimation process. In this paper, we propose a model-data dual-driven (MDD) scheme to jointly perform the model-driven non-stationary channel denoising and the data-driven channel interpolation in an end-to-end way, which is followed by a low-complexity channel refinement module to improve the robustness of the proposed scheme. Specifically, image contour extraction (ICE) is utilized to effectively eliminate the non-stationary noises in the channel matrices before being sent to the downstream interpolation network. An enhanced convolutional neural network (CNN)-based residual network (eCNN-RN) is developed to perform non-linear interpolations for recovering the U-MIMO-OFDM channels. Based on ICE, the proposed online refinement module can improve the generalizability of the learned model to a practical environment. Numerical experiments demonstrate the efficiency and the effectiveness of the cross-fertilization of the model-driven and data-driven approaches.

A cross-domain iterative receiver structure is proposed for orthogonal time frequency space (OTFS) systems over sparse doubly selective channels. It achieves higher reliability by alternatively performing channel estimation and signal detection across time-frequency (TF) domain and delay-Doppler (DD) domain. In particular, with the aid of some hybrid pilots designed carefully, TF-domain channel estimation enables higher accuracy than the existing DD-domain channel estimation schemes, thanks to less interference from the data symbols in TF domain. Low-complexity data detection is performed in DD domain to exploit the full diversity of OTFS. Furthermore, the detected data is removed from the received signal to further eliminate interference and enhance channel estimation. Simulation results demonstrate that the proposed cross-domain iterative OTFS receiver outperforms the existing schemes in terms of mean square error of channel estimation and bit error rate, with less pilot overhead.

Bayesian modelling of the channel distribution is a crucial step before channel recovery specially in the underdetermined scenario in multiple input multiple output (MIMO) antenna setups. In complicated dynamic propagation environments such as the ones encountered in Unmanned Aerial Vehicles (UAVs) Air to Ground (A2G) channels, Bayesian modelling might not be feasible or the model may not be able to approximate the different aspects of the true distribution well enough. Thus, estimation performance will be affected irrespective of the efficiency of recovery algorithm. To exploit the temporal correlations and imperfections in the real channels in such a scenario, we design a temporally correlated adversarial regulariser using Variational recurrent neural networks (VRNN) and train the framework on simulated channel dataset. The framework can be trained directly with channel samples, thus, allowing channel modelling and estimation without explicit tractable Bayesian models in highly dynamic systems. We then propose a temporally correlated deep compressed sensing algorithm which does not depend on the expressibility of the networks and provide theoretical results for existence and recovery. Numerical experiments demonstrate its effectiveness for channel estimation in A2G channels and show superior channel recovery and improved modelling even for out-of-distribution channels.

Most existing downlink channel estimation methods rely on channel sparsity in the angular domain to reduce pilot overhead for massive multiple-input multiple-output (MIMO) systems. Compared with channel estimation methods based on predefined basis or offline dictionary learning, in this paper, we design an online two-timescale joint uplink/downlink dictionary learning and channel estimation (TTS-JDLCE) algorithm for dynamic scenarios, where the channel sparsity and angle reciprocity between uplink and downlink transmissions are both exploited to reduce the pilot overhead. The downlink channel estimation is constructed as a TTS stochastic optimization problem with a constraint coupled by the long-term dictionary and short-term sparse channel representations. Treating the dictionary as a learnable parameter, the proposed algorithm can capture dynamic spatial information for enhancing performance. By introducing a relaxed TTS primal-dual decomposition (PDD) framework, the original problem is decomposed into a long-term online dictionary learning subproblem and a family of short-term sparse channel estimation subproblems. Besides, the deep unfolding technique is employed to extract gradient information from short-term problems, which circumvents the non-closed form and non-convexity of long-term subproblem by constructing a convex surrogate problem. Finally, simulations show that the proposed method remarkably reduces the pilot overhead and achieves significant performance gains over various baselines.

Short frame structure and its optimization plays an important role in ultra reliable and low latency communication (URLLC). We investigate and compare the latency and throughput performances of the independent encoding (IE) and joint encoding (JE) frame structures for heterogeneous multi-device URLLC in the finite block length regime. There is a counter-intuitive finding that, despite a longer frame, IE enables a much lower average latency and higher reliability than JE, thanks to lower queuing latency, while JE achieves higher throughput with lower traffic heterogeneity, thanks to less channel dispersion. It is also shown that traffic heterogeneity has less adverse effects on the performance of the IE frame structure, and can even help reduce its average latency with the shortest block length first (SBF) scheduling rule proposed. We also provide an intensive analysis of the trade-off between pilot power and pilot overhead, with near-optimal pilot power and block length derived in closed form. Low-complexity joint pilot power, pilot length and block length optimization algorithms are proposed for IE and JE frame structures. Numerical results verify the effectiveness of the proposed algorithms, and also show that pilot power optimization plays a significant role in enhancing throughput at low to medium SNR.

Efficient channel state information (CSI) compression and feedback from user equipment to the base station (BS) are crucial for achieving the promised capacity gains in massive multiple-input multiple-output (MIMO) systems. Deep autoencoder (AE)-based schemes have been proposed to improve the efficiency of CSI compression and feedback. However, existing AE-based schemes suffer from critical issues in both CSI dimensionality reduction and latent feature quantization. In this paper, we propose a novel hierarchical sparse AE for efficient CSI compression and feedback for the 5G-NR fixed-length CSI feedback mechanism. Our approach employs a two-tier AE structure to jointly compress the sparse CSI latent feature and its side information. Additionally, we utilize a model-assisted Bayesian Rate-Distortion approach to train the weights of the AE. Specifically, the training loss function is formulated based on the variational Bayesian inference framework given a parametric Bernoulli Laplace Mixture prior model and a sparsity-inducing likelihood model. Furthermore, we propose a model-assisted adaptive coding algorithm to quantize the latent feature under the fixed codeword bit length constraint. Our experimental results demonstrate that the proposed solution outperforms existing AE-based schemes under various feedback budgets.

Over-the-Air aggregation (OTA) is a promising technology for Internet-of-Things (IoT) applications, but it can be vulnerable to burst interference from co-channel non-cooperative IoT communications. In this paper, we propose a robust OTA framework for IoT-OTA applications that mitigates burst sparse interference using analog product code. We first propose a hierarchical likelihood model and a Markov-based hierarchical prior model to capture the structural properties in the received coded observations and the statistics of the burst-sparse interference, respectively. Then, we design a low-complexity Bayesian decoder using online turbo variational Bayesian inference technique, which ensures good performance without prior knowledge of the model parameters. In addition, we propose an optimization-based design for the analog product code encoding matrix that minimizes the end-to-end decoding mean squared error. Extensive simulations demonstrate the effectiveness of the proposed solution on interference mitigation, outperforming various state-of-the-art baseline schemes.

A sparse superimposed vector transmission (SSVT) scheme is proposed for short-packet ultra-reliable and low-latency communications (URLLC) in high-mobility scenarios. Unlike conventional pilot-less sparse vector coding (SVC) schemes, a small number of pilots are superimposed on top of data vector in SSVT to track time-varying channels, leading to high spectral efficiency. Two joint channel estimation and data decoding algorithms, called iterative interference cancellation (IIC) and iterative data-aided (IDA), are proposed to recover data from the received signal. The IIC algorithm accurately decodes data by iteratively canceling pilot and data interference, while the IDA algorithm further refines channel estimation by treating previously decoded data as pseudo-pilots. Simulation results show that the proposed SSVT scheme significantly outperforms conventional SVC schemes in terms of block error rate (BLER), and the proposed IIC and IDA algorithms show fast convergence speed.

In this letter, we investigate the robust beamforming design for a hybrid active-passive reconfigurable intelligent surface (RIS) assisted integrated sensing and communication (ISAC) system, where the RIS consists of both active and passive reflecting elements. Different from the existing works, we consider the target location estimation error. We aim to maximize the minimum sensing beampattern gain among multiple targets, subject to the signal-to-interference-plus-noise ratio (SINR) constraint for each communication user (CU), the transmit power constraints for the BS and the active elements. To address the non-convexity of the max-min RIS-assisted ISAC design problem, we propose an alternating-optimization based design solution to alternately optimize the BS’s transmit beamforming design and the hybrid RIS’s coefficient matrix in a low-complexity manner. Simulation results demonstrate the benefit of the proposed design scheme over the existing baseline schemes.

We consider a multiple-input single-output (MISO) optical wireless communications (OWC) system with highly correlated channels, causing bit error rate (BER) performance degradation. Because of intensity modulation and direct detection (IM/DD), the transmitted signals are real-valued and non-negative, which limits the utilization of precoding in phase domain. We employ direct current biased optical orthogonal frequency division multiplexing (DCO-OFDM) modulation to design a group of phase rotation (PR) factors in frequency domain for OWC systems, robust against high channel correlations. The proposed PR-based precoding has a number of advantages over the power factor-based design in the literature: i) no change of transmission power on each LED; ii) no signal-to-noise ratio (SNR) and no BER degradation on transmitted signals of all LEDs. We formulate an optimization problem to obtain the optimal PR factors by maximizing the minimum pairwise Euclidean distances between all candidate signals. The optimization problem is non-convex, requiring multi-dimensional exhaustive searches. In order to reduce the complexity, we propose three low-complexity PR-based precoding approaches which provide BER performances better than the power factor-based designs in the literature, and are close to their own analytical results derived, respectively. The proposed approaches are validated via a built testbed, producing comparable performance between measured and simulated BERs.

In this article, we consider identification and control for potentially unstable linear systems with multiagent controllers in the presence of wireless interference channels among the remote controllers and the actuators. We formulate the optimal control problem as a stochastic game and utilize the stochastic approximation technique to learn the optimal control solutions and the plant dynamics simultaneously in an online manner. Using Lyapunov drift analysis, we show that the identified plant dynamics and the learned control solutions via the proposed algorithm can converge to the true plant dynamics and optimal control solutions, respectively. Furthermore, we also rigorously show the convergence speed of the proposed learning algorithm in terms of the online learning step sizes in the proposed algorithm. The proposed scheme is also compared with various state-of-art baselines and we show that significant performance gains can be achieved.

Despite the many advantages of using deep neural networks over shallow networks in various machine learning tasks, their effectiveness is compromised in a federated learning setting due to large storage sizes and high computational resource requirements for training. A large model size can potentially require infeasible amounts of data to be transmitted between the server and clients for training. To address these issues, we investigate the traditional and novel compression techniques to construct sparse models from dense networks whose storage and bandwidth requirements are significantly lower. We do this by separately considering compression techniques for the server model to address downstream communication and the client models to address upstream communication. Both of these play a crucial role in developing and maintaining sparsity across communication cycles. We empirically demonstrate the efficacy of the proposed schemes by testing their performance on standard datasets and verify that they outperform various state-of-the-art baseline schemes in terms of accuracy and communication volume.

Dictionary learning is a widely used unsupervised learning method in signal processing and machine learning. Most existing works on dictionary learning adopt an off-line approach, and there are two main off-line ways of conducting it. One is to alternately optimize both the dictionary and the sparse code, while the other is to optimize the dictionary by restricting it over the orthogonal group. The latter, called orthogonal dictionary learning (ODL), has a lower implementation complexity and, hence, is more favorable for low-cost devices. However, existing schemes for ODL only work with batch data and cannot be implemented online, making them inapplicable for real-time applications. This article, thus, proposes a novel online orthogonal dictionary scheme to dynamically learn the dictionary from streaming data, without storing the historical data. The proposed scheme includes a novel problem formulation and an efficient online algorithm design with convergence analysis. In the problem formulation, we relax the orthogonal constraint to enable an efficient online algorithm. We then propose the design of a new Frank–Wolfe-based online algorithm with a convergence rate of $\mathcal {O}(\ln t/t^{1/4})$ . The convergence rate in terms of key system parameters is also derived. Experiments with synthetic data and real-world Internet of things (IoT) sensor readings demonstrate the effectiveness and efficiency of the proposed online ODL scheme.

In this paper, we investigate an intelligent reflecting surface (IRS)-assisted integrated sensing and communication (ISAC) system design in a clutter environment. Assisted by an IRS equipped with a uniform planar array (UPA), a multi-antenna base station (BS) is targeted for simultaneously sensing multiple targets in the non-light-of-sight (NLoS) region and communicating with multiple communication users (CUs). We consider the joint IRS-assisted ISAC design in the case with Type-I or Type-II CUs, where each Type-I CU and Type-II CU can and cannot cancel the interference from sensing signals, respectively. Under the perfect communication/sensing channel state information assumption, we aim to maximize the minimum sensing beampattern gain among multiple targets, where the sensing beampattern gain qualifies the achieved illumination signal power at the given location of the target of interest. We jointly optimize the BS’s communication-sensing beamformers and the IRS’s phase shifting matrix, subject to the signal-to-interference-plus-noise ratio (SINR) constraint for each Type-I/Type-II CU, the interference power constraint per clutter, the transmission power constraint at the BS, and the cross-correlation pattern constraint. Due to the design variable coupling, the joint IRS-assisted ISAC design problem is shown to be non-convex in the case with Type-I or Type-II CUs. To circumvent the non-convexity dilemma, we propose semidefinite relaxation (SDR) based alternating optimization algorithms in both cases, for which the computational complexity and convergency behavior are analyzed. In the case with Type-I CUs, we show that the dedicated sensing signal at the BS can help enhance the sensing performance gain. By contrast, the dedicated sensing signal at the BS is not required for the IRS-assisted ISAC designs in the case with Type-II CUs. Numerical results are provided to show that the proposed IRS-assisted ISAC design schemes achieve a significant gain over the existing benchmark schemes.

Multiplexing or transmission diversity via dual links enables the reduction in age of information (AoI). We address the open issue of selection between the two transmission modes for AoI-oriented short-packet system over fading channels, with a comprehensive analysis. Closed-form expressions for the average AoI and peak AoI (PAoI) are derived based on the discrete-time Markov-chain process. Then, to obtain the explicit region of preference (RoP) and quantitative PAoI gains of multiplexing/diversity over the single-queue case, we derive the signal-to-noise ratio (SNR) threshold for transmission mode selection, which is shown to be a decreasing function of the arrival rate and saturates at high arrival rate. Also, the monotonicity of the PAoI gains by multiplexing and diversity, and their achievable gains, are analyzed in a comprehensive manner. It is shown that diversity is able to achieve a PAoI gain of more than 3 dB over the single-queue case at low SNR, while multiplexing has a larger RoP than diversity and is selected at high SNR and high arrival rate. Furthermore, both throughput and PAoI violation probability are considered alongside the average PAoI for a wide range of tradeoff in system design.

This letter investigates a cache-enabled multiuser mobile edge computing (MEC) system with dynamic task arrivals, taking into account the impact of proactive cache placement on the system’s overall energy consumption. We consider that an access point (AP) schedules a wireless device (WD) to offload computational tasks while executing the tasks of a finite library in the task caching phase, such that the nearby WDs with the same task request arriving later can directly download the task results in the task arrival and execution phase. We aim for minimizing the system’s weighted-sum energy over a finite-time horizon, by jointly optimizing the task caching decision and the MEC execution of the AP, and local computing as well as task offloading of the WDs at each time slot, subject to caching capacity, task causality, and completion deadline constraints. The formulated design problem is a mixed-integer nonlinear program. Under the assumption of fully predicable task arrivals, we first propose a branch-and-bound (BnB) based method to obtain the optimal offline solution. Next, we propose two low-complexity schemes based on convex relaxation and task-popularity, respectively. Finally, numerical results show the benefit of the proposed schemes over existing benchmark schemes.

In this article, we consider the optimal control of linear systems over wireless multiple-input multiple-output (MIMO) fading channels, where the wireless MIMO fading and random access of the remote controller may cause intermittent controllability or uncontrollability of the closed-loop control system. We consider the finite-time horizon optimal control first and then check the conditions for the existence and uniqueness of optimal control for the infinite-time horizon case. Specifically, for the finite-time horizon case, we show that the optimal control gain matrix can be computed offline based on the distribution of the wireless MIMO fading and random access of the controller. For the infinite-time horizon case, we show that, when the closed-loop system is almost surely controllable, the optimal control solution always exists and is unique. In the case that MIMO fading channels and the random access of the remote controller destroy the closed-loop controllability, we propose a novel controllable and uncontrollable positive semidefinite cone decomposition induced by the singular value decomposition of the MIMO fading channel contaminated control input matrix. This enables a closed-form characterization of the sufficient condition for both the existence and the uniqueness of the optimal control solution. The closed-form sufficient condition reveals the fact that the optimal control action may still exist even if the closed-loop system suffers from intermittent controllability or almost sure uncontrollability. We further propose a novel stochastic approximation (SA)-based online learning algorithm that can learn the optimal control action on the fly based on the plant state observations. We finally extend the technical results to the case in which the channel state information (CSI) is unavailable at the controller. We show that the sufficient condition for the existence and uniqueness of the optimal control solution is more difficult to satisfy due to the penalty caused by the unknown CSI. The proposed scheme is also compared with various baselines, and we show that significant performance gains can be achieved.

To alleviate the local computation demands from the ever-increasing computation-intensive mobile applications, Mobile Edge Computing (MEC) has proved promising. Especially, by opportunistically offloading these computation tasks to the MEC server, the delay of computing could be significantly improved through communication. In this paper, we develop an analytical framework for joint communication and computation resources allocation for multi-user MEC systems. Specifically, to retrieve the combined effect of communication and computation capabilities, we establish a dual queue system, including a data queue sub-system and a computation queue sub-system. To address the associated stochastic resource optimization problem, we propose a low-complexity resource allocation algorithm by Lyapunov optimization to stabilize all the sub-queue systems. As the practical buffers are finite, the conventional delay analysis of Lyapunov optimization becomes inaccurate. Alternatively, we model the stochastic queue lengthes as discrete time controlled random walk processes, which are transformed to continuous time Stochastic Differential Equations (SDEs) with reflections by strong approximation. According to the steady state analysis on the SDEs, we derive closed-form steady state distributions of the queue lengths, and then obtain the average delay performance with finite buffers. Finally, the accuracy of the proposed delay analysis is verified through simulation.

Deep learning (DL)-based channel estimation has achieved remarkable success. However, most existing works focus on the white Gaussian noise which are inapplicable for cell-edge users under inter-cell interference (ICI). In this paper, we address this issue by proposing a novel DL-based channel estimation solution with a cascaded model-based and model-free deep neural network (DNN) structure. Specifically, the model-based module is designed by the variational Bayesian inference (VBI) technique to suppress the time-varying ICI, and the model-free module is designed by the Denoising Sparse Autoencoder (DSAE) structure to further refine the channel estimation. The proposed DNN is firstly pre-trained by offline supervised training, and various channel statistics are encapsulated in the DNN weights with the assist of a hyper-prior net modelling different sparse priors for different training samples. Then, an online Bayesian learning algorithm is proposed to train the model-based VBI module based on real-time pilot samples to track the online channel statistics. Simulation results verify that the proposed solution outperforms various state-of-the-art baseline schemes in a large SINR range with comparable performance to the estimator with genie-aided channel statistics.

In this paper, we study a decentralized deep neural network (DNN) task partitioning and offloading control problem for a multi-access edge computing (MEC) system with multiple wireless devices (WDs) powered by renewable energy sources. Instead of generic computational tasks, we focus on the case when the WDs and the MEC systems are computing DNN inferencing tasks. For each WD, we build a virtual local-computing CPU workload queue and an energy queue to model the DNN computation dynamics and the battery energy dynamics, respectively. We first introduce a new Lyapunov function based on the DNN tasks' cumulative execution latency and the total energy consumptions of the entire MEC system. Using a Lyapunov optimization approach, the DNN task partitioning and offloading control at the WDs are formulated as a Lyapunov drift minimization problem. To facilitate a distributed implementation, we exploit the linear structure of the cumulative latency and the independency of each WD's local-computing CPU rate, and we decompose the centralized Lyapunov drift minimization problem into multiple distributed subproblems, each of which is associated with a single WD. Next, we develop a parametric online learning algorithm such that each sub-Lyapunov drift minimization problem is solved locally at its associated WD. The proposed solution is completely decentralized in the sense that the sequential offloading and DNN task segmentation control at each WD is determined by its local battery energy queue state information (EQSI), the channel state information (CSI), and the DNN inferencing task workload queue state information (TQSI). Numerical results reveal that the proposed online learning algorithm can achieve significant performance gain over various state-of-the-art baselines.

With the growth of neural network size, model compression has attracted increasing interest in recent research. As one of the most common techniques, pruning has been studied for a long time. By exploiting the structured sparsity of the neural network, existing methods can prune neurons instead of individual weights. However, in most existing pruning methods, surviving neurons are randomly connected in the neural network without any structure, and the non-zero weights within each neuron are also randomly distributed. Such irregular sparse structure can cause very high control overhead and irregular memory access for the hardware and even increase the neural network computational complexity. In this paper, we propose a three-layer hierarchical prior to promote a more regular sparse structure during pruning. The proposed three-layer hierarchical prior can achieve per-neuron weight-level structured sparsity and neuron-level structured sparsity. We derive an efficient Turbo-variational Bayesian inferencing (Turbo-VBI) algorithm to solve the resulting model compression problem with the proposed prior. The proposed Turbo-VBI algorithm has low complexity and can support more general priors than existing model compression algorithms. Simulation results show that our proposed algorithm can promote a more regular structure in the pruned neural networks while achieving even better performance in terms of compression rate and inferencing accuracy compared with the baselines.

Industrial metaverse aims to empower holographic intelligent manufacturing by mapping the industrial environment into a virtual world in real time. Wireless communication plays a pivotal role as a bridge between the real and virtual worlds, making it indispensable for industrial metaverse. However, limited radio resources, combined with diverse requirements of various tasks, bring significant challenges in achieving satisfactory data transmission in industrial metaverse. In this context, we are motivated to investigate intelligent wireless communication, i.e., task-oriented communication designated to meet the specific requirements of three main types of tasks, for industrial metaverse. First, we explore high timeliness semantic communication for real world virtualization tasks to fulfill the vast data transmission requirements of massive number of devices. We then discuss the closed-loop communication tailored for virtual world realization to tackle the challenges brought by deeply inter-connected uplink and downlink. For cooperative model training tasks, federated learning with data evaluation and limited information sharing is investigated to address the concerns about privacy breaches and scarcity of radio resources. Finally, we discuss open challenges of task-oriented communication for industrial metaverse.

Over-the-air computation (AirComp) has emerged as a promising technique in future intelligent wireless networks, which enables swift functional computation among distributed wireless devices (WDs) by exploiting the superposition property of wireless channels. This paper studies a new hierarchical AirComp network over a large area, in which a set of intermediate relays are exploited to facilitate the massive data aggregation from a large number of WDs. Under this setup, we present a two-phase amplify-and-forward (AF) relaying protocol. In the first phase, the WDs simultaneously send their data to the relays, while in the second phase, the relays amplify the respectively received signals and concurrently forward them to the fusion center (FC) for aggregation. Our objective is to minimize the computational mean squared error (MSE) at the FC, by jointly optimizing the transmit coefficients of the WDs, the AF coefficients of the relays, and the de-noising factor of the FC, subject to their individual transmit power constraints. First, we consider the centralized design with global channel state information (CSI), in which the inter-relay signals can be exploited beneficially for data aggregation. In this case, we develop an alternating-optimization-based algorithm to obtain a high-quality solution to the computational MSE minimization problem. The obtained solution shows that the phase of the transmit coefficient at each WD is opposite to that of the WD-relay-FC channel to ensure the signal phase alignment at the FC, and the transmit power of each WD/relay follows a regularized composite-channel-inversion structure to strike a balance between minimizing the signal-magnitude-misalignment-induced error and the noise-induced error. Next, to reduce the signaling overhead caused by the centralized design, we consider an alternative decentralized design with partial CSI, in which the relays and the FC make their own decisions by only requiring the channel power gain information across different relays. In this case, the relays and FC need to treat the inter-relay signals as harmful interference or noise. Accordingly, we optimize the transmit coefficients of the WDs associated with each relay, and the relay AF coefficients (together with the FC de-noising factor) in an iterative manner, which can be implemented efficiently in a decentralized way. Finally, numerical results show the fast convergence of the proposed centralized and decentralized designs. It is also shown that both designs achieve significant MSE performance gains over benchmark schemes without the joint optimization.

In this paper, reinforcement learning (RL) based end-to-end channel prediction (CP) and beamforming (BF) algorithms are proposed for multi-user downlink system. Different from the previous methods which either require perfect channel state information (CSI), or estimate outdated CSI and set constraints on pilot sequences, the proposed algorithms have no such premised assumptions or constraints. Firstly, RL is considered in channel prediction and the actor-critic aided CP algorithm is proposed at the base station (BS). With the received pilot signals and partial feedback information, the actor network at BS directly outputs the predicted downlink CSI without channel reciprocity. After obtaining the CSI, BS generates the beamforming matrix using zero-forcing (ZF). Secondly, we further develop a deep RL based two-layer architecture for joint CP and BF design. The first layer predicts the downlink CSI with the similar actor network as in the CP algorithm. Then, by importing the outputs of the first layer as inputs, the second layer is the actor-critic based beamforming layer, which can autonomously learn the beamforming policy with the objective of maximizing the transmission sum rate. Since the learning state and action spaces in the considered CP and BF problems are continuous, we employ the actor-critic method to deal with the continuous outputs. Empirical numerical simulations and the complexity analysis verify that the proposed end-to-end algorithms could always converge to stable states under different channel statistics and scenarios, and can beat the existing traditional and learning based benchmarks, in terms of transmission sum rate.

In this article, we consider a multiagent linear control system, where multiple actuators are tightly coupled with an unstable dynamic plant and connected with a couple of geographically scattered remote controllers over wireless interference channels. We propose a novel decentralized dynamic scheduling and control scheme for multiple control agents, where the associated solutions are fully decentralized with an event-triggered structure in the sense that the structure depends on the instantaneous plant state and local channel state for each controller. We establish the closed-form sufficient requirements on the communication resources needed to achieve stabilization of the multiagent system using the Lyapunov drift analysis approach. Numerical results show that our proposed scheme can achieve significant performance gain over various state-of-the-art baselines for scheduling and control of the multiagent system.

In this article, we consider the remote state estimation for nonlinear dynamic systems with known linear dynamics and unknown nonlinear perturbations. The nonlinear dynamic plant is monitored by multiple distributed sensors over a random access wireless network with shared common radio channel. We focus on the communication strategy and remote state estimation algorithm design so as to achieve a remote state estimation stability subject to unknown nonlinearities in plant and various wireless impairments, such as multisensor interference, wireless fading, and additive channel noise. By exploiting the additive properties of the physical wireless channels, we propose a novel information fusion over-the-air mechanism to address the signal collision and interference among the sensors. Utilizing the partial knowledge on the linear dynamics of the plant, we also propose a novel recurrent neural network (RNN)-based remote state estimator aided by a virtual state estimation mean-square-error (MSE) process. We further propose a novel online training algorithm such that the RNN at the remote estimator can effectively learn the unknown plant nonlinearities. Using the Lyapunov drift analysis approach, we establish closed-form sufficient requirements on the communication resources needed to achieve almost sure stability of both state estimation and RNN online training in high signal-to-noise ratio (SNR) regime. As a result, our proposed scheme is asymptomatic optimal for large SNR in the sense that both the plant state and the unknown plant nonlinearities can be perfectly recovered at the remote estimator. The proposed scheme is also compared with various baselines and we show that significant performance gains can be achieved.

In this paper, we consider a cyber-physical system, where an unstable dynamic plant is monitored by multiple distributed sensors over a wireless communication network. We propose a systematic stability analysis framework for multisensor cyber-physical systems. The proposed framework provides closed-form characterizations of a stability condition for general measurement matrices with rank-deficient cases covered. Utilizing the observable and unobservable cone decomposition of the coordinate transformed system induced by the singular value decomposition of the measurement matrices, we decompose the conventional Riccati recursion of state estimation covariance. Based on the decomposed recursion, we establish closed-form sufficient requirements on the communication resources needed to achieve stabilization of multisensor cyber-physical systems using the Lyapunov drift analysis approach. The proposed framework is also compared with various representative literature and we show that significant performance gains can be achieved.

Massive multiple input multiple output (MIMO) is one of the promising technologies to address the growing capacity requirement in fifth-generation cellular systems. However, the massive MIMO deployed at base stations (BSs) or relay stations (RSs) brings new challenges on channel state information (CSI) feedback because of its large-scale complex channel matrix. In this paper, we propose a hybrid limited feedback with selective eigenvalue information (HLFSEI), which adopts the individual quantized feedback and the codebook-based feedback jointly. In HLFSEI, we feed back only selective eigenvalue elements for power allocation by individual quantized feedback and the precoding matrix by codebook-based feedback. Furthermore, we study the optimal feedback bit allocation of the aforementioned two feedback methods to minimize the throughput loss with the feedback-link capacity constraint. Specifically, the feedback bits are allocated according to the properties of the throughput loss in high- and low-signal-to-noise regimes. Both the BS–RS communications and the BS/RS–user-equipment communications in cellular systems are taken into consideration. Finally, we evaluate the performance of the proposed HLFSEI strategy by simulation and show its performance gain compared with conventional feedback strategies.

Future smart grids will be featured by flexible supply-demand management and great penetration of renewable energy to enable efficient and economical grid operations. While the renewable energy offers a cheaper and cleaner energy supply, it introduces supply uncertainty due to the volatility of renewable source. It is therefore of practical importance to investigate the optimal exploitation of renewable energy based on the supply-demand framework of future smart grids, where the energy-providers (or the energy-users) adaptively adjust their energy-provisioning (or energy-demands) according to some system state information that takes into account the volatility of renewable energy. Specifically, we consider cost-efficient energy scheduling for residential smart grids equipped with a centralized renewable energy source. Our scheduling problem aims at: 1) quantifying the optimal utilization of renewable energy that achieves the tradeoff between the system-wide benefit from exploiting the renewable energy and the associated cost due to its volatility; and 2) evaluating how the volatility of renewable energy influences its optimal exploitation. We also propose computationally efficient and distributed algorithms to determine the optimal exploitation of renewable energy as well as the associated energy scheduling decisions.

Turbo‐coded Vertical Bell Labs Layered Space‐Time Architecture (V‐BLAST) is a multiple‐antenna system employing turbo codes and bit‐interleaving to offer flexible rates and high‐speed wireless data communications. The conventional detection techniques based on one‐shot demodulation and decoding algorithms may suffer poor performance or simply fail when there are fewer receive antennas than transmit antennas or when the multiple‐input‐multiple‐output (MIMO) channel exhibits strong spatial correlations. We propose employing an iterative demodulation and decoding algorithm that solves these problems and greatly extends the applicability of V‐BLAST. Simulations demonstrate that iterative V‐BLAST offers the best‐known performance in a wide range of settings. Through complexity analysis, we show that (a) the incremental computing complexity is small, although additional memory is required, compared with the conventional one‐shot maximum likelihood (ML) decoding method, and (b) an iterative algorithm employing generalized sphere decoding will be a low‐complexity, memory‐reduced, and high‐performance solution. We also give extensions to the technique beyond simple V‐BLAST. In addition, when feedback from the receiver to the transmitter is available, we illustrate how to adapt code rate, modulation, and the power distribution to take advantage of the channel knowledge. With a combination of the “best in class” receiver performance and the ability to decode V‐BLAST transmissions with fewer receive antennas than transmit antennas, the techniques described here can be viewed as key “technology enablers” for BLAST/MIMO in next‐generation systems. © 2003 Lucent Technologies Inc.

The multi‐user multiple‐input‐multiple‐output (MIMO) space–time scheduling problem is the prime focus of the paper. Using multiple transmit (nT) and receive antennas (nR), it is well known that the link‐level throughput will be increased linearly with respect to min[nT, nR] without increasing the bandwidth and power budget. However, optimizing the link‐level performance of the MIMO system does not always imply achieving scheduling‐level optimization. Therefore, the design optimization across the link layer and the scheduling layer is very important to fully exploit the temporal and spatial dimensions of the communication channel. In this paper, we address the design of the optimal space–time scheduler for the multi‐user MIMO system based on an information theoretical approach. Since full knowledge of channel matrix at the transmitter requires a feedback channel capacity not scalable with respect to nT (number of transmit antennas at mobiles) and nR (number of receive antennas at base station), we will assume two partial feedback models. © 2003 Lucent Technologies Inc.

In this paper, we introduce an adaptive coding scheme for a direct‐sequence code division multiple access (DS‐CDMA) system. The system uses noncoherent M‐ary orthogonal modulation with a RAKE receiver and power control. We considered both a fast fading channel and a combined fast fading, shadowing, and power control channel. To evaluate the performance of the system, we derived analytical bounds and conducted simulations. We found significant improvements in the average throughput and the bit error rate (BER) performance using the adaptive coding scheme. The amount of improvement drops as the number of diversity branches increases. More importantly, we found that the adaptive coding scheme is relatively robust to shadowing, while fixed‐rate codes are ineffective in the shadowing environment. The adaptive coding scheme is also robust to mobile speed, feedback delay, and finite interleaving depth.