This paper studies an uplink dual-functional sensing and communication system aided by a reconfigurable intelligent surface (RIS), whose reflection pattern is optimally configured to trade-off sensing and communication functionalities. Specifically, the Bayesian Cramér-Rao lower bound (BCRLB) for estimating the azimuth angle of a sensing user is minimized while ensuring the signal-to-interference-plus-noise ratio constraints for communication users. We show that this problem can be formulated as a novel fractionally constrained fractional programming (FCFP) problem. To deal with this highly nontrivial problem, we extend a quadratic transform technique, originally proposed to handle optimization problems containing fractional structures only in objectives, to the scenario where the constraints also include ratios. First, we consider the case where the fading coefficient is known. Using the quadratic transform, the FCFP problem can be turned into a sequence of subproblems that are convex except for the constant-modulus constraints which can be tackled using a penalty-based approach. To further reduce the computational complexity, we leverage the constant-modulus conditions and propose a novel linear transform. This new transform enables the FCFP problem to be turned into a sequence of linear programming (LP) subproblems, which can be solved with linear complexity in the dimension of reflecting elements. Then, we consider the case where the fading coefficient is unknown. A modified BCRLB is used to make the problem more tractable, and the proposed quadratic transform-based algorithm is used to solve the problem. Numerical results unveil nontrivial and effective reflection patterns that can be synthesized by the RIS to facilitate both communication and sensing functionalities.

This paper addresses the integration of millimeter-wave (mmWave) Wi-Fi communication and sensing during data transmission intervals (DTIs). We leverage prior knowledge from codebook beam training conducted during preceding beacon transmission intervals (BTIs) and association beamforming training (A-BFT) intervals to design a transceiver array response that meets both requirements on downlink communication SNR and targeted sensing area. By formulating it as a first-order array response optimization with constraints on power, codebook, communication SNR, and limited RF chains, this paper introduces a two-stage solution. First, we introduce a two-way communication-sensing matching pursuit to determine a set of codewords that prioritize the communication SNR constraint. Then, using the selected codewords, we employ an alternating minimization over an auxiliary phase term and beamforming weights to further minimize an array-response distance loss. Numerical results validate the effectiveness of the proposed DMG beamforming design over baseline methods.

This paper considers a massive connectivity setting in which a base-station (BS) aims to communicate sources (X1, · · · ,Xk) to a randomly activated subset ofkusers, among a large pool ofnusers, via a common message in the downlink. Although the identities of thekactive users are assumed to be known at the BS, each active user only knows whether itself is active and does not know the identities of the other active users. A naive coding strategy is to transmit the sources alongside the identities of the users for which the source information is intended. This requiresH(X1, · · · ,Xk) +klog(n) bits, because the cost of specifying the identity of one out ofnusers is log(n) bits. For largen, this overhead can be significant. This paper shows that it is possible to develop coding techniques that eliminate the dependency of the overhead onn, if the source distribution follows certain symmetry. Specifically, if the source distribution is independently and identically distributed (i.i.d.) then the overhead can be reduced to at mostO(log(k)) bits, and in case of uniform i.i.d. sources, the overhead can be further reduced toO(1) bits. For sources that follow a more general exchangeable distribution, the overhead is at mostO(k)bits, and in case of finite-alphabet exchangeable sources, the overhead can be further reduced toO(log(k)) bits. The downlink massive random access problem is closely connected to the study of finite exchangeable sequences. The proposed coding strategy allows bounds on the Kullback-Leibler (KL) divergence between finite exchangeable distributions and i.i.d. mixture distributions to be developed, and gives a new KL divergence version of the finite de Finetti theorem which is scaling optimal.

This paper presents a novel optimization framework for beamforming design in integrated sensing and communication systems where a base station seeks to minimize the Bayesian Cramer-Rao bound of a sensing problem while satisfying quality of service constraints for the communication users. Prior approaches formulate the design problem as a semidefinite program for which acquiring a beamforming solution is computationally expensive. In this work, we show that the computational burden can be considerably alleviated. To achieve this, we transform the design problem to a tractable form that not only provides a new understanding of Cramer-Rao bound optimization, but also allows for an uplink-downlink duality relation to be developed. Such a duality result gives rise to an efficient algorithm that enables the beamforming design problem to be solved at a much lower complexity as compared to the-state-of-the-art methods.

This paper considers a massive connectivity scenario in which a base-station (BS) aims to communicate k individual sources (X1, ⋯ , Xk) to a random subset of k users among a large pool of n users via a common downlink message. The identities of the k active users are known at the BS, but each active user only knows whether it is active itself and does not know the identities of the other active users. The naive coding strategy of transmitting the source messages together with the indices of the users for which the messages are intended would require a rate of H(X1, ⋯ , Xk) + k log(n) bits. This paper shows that if the sources are jointly distributed according to an exchangeable distribution, better coding techniques can be used to eliminate the dependency of the overhead on log(n). Specifically, if the sources are independently and identically distributed (i.i.d.) or are i.i.d. mixture, then the overhead can be reduced to O(log(H(X1, ⋯ , Xk))) or at most O(log(k)) bits. The overhead can be further reduced to O(1) if the source distribution is uniform over its support. For a general exchangeable source not necessarily i.i.d. nor i.i.d. mixture, an overhead of O(k + log(k + H(X1, ⋯, Xk))) bits is achievable; if the source distribution has finite support, the overhead can be further reduced to O(log(k)). Moreover, for exchangeable distributions that are extendable, the rate can be further improved.

Balancing traffic among cellular networks is very challenging due to many factors. Nevertheless, the explosive growth of mobile data traffic necessitates addressing this problem. Due to the problem complexity, data-driven self-optimized load balancing techniques are leading contenders. In this work, we propose a comprehensive deep reinforcement learning (RL) framework for steering the cell individual offset (CIO) as a means for mobility load management. The state of the LTE network is represented via a subset of key performance indicators (KPIs), all of which are readily available to network operators. We provide a diverse set of reward functions to satisfy the operators' needs. For a small number of cells, we propose using a deep Q-learning technique. We then introduce various enhancements to the vanilla deep Q-learning to reduce bias and generalization errors. Next, we propose the use of actor-critic RL methods, including Deep Deterministic Policy Gradient (DDPG) and twin delayed deep deterministic policy gradient (TD3) schemes, for optimizing CIOs for a large number of cells. We provide extensive simulation results to assess the efficacy of our methods. Our results show substantial improvements in terms of downlink throughput and non-blocked users at the expense of negligible channel quality degradation.

This paper considers a massive random access scenario in which a small set of k users out of a large number of n potential users are active at any given time, and a central base-station wishes to send a common message to the active users in order to label them into a finite number of categories. Specifically, given c possible categories, the base-station wishes to send label ℓ to a set of kℓusers, where ℓ ∈ {1, …, c} and $\sum\nolimits_{\ell = 1}^c {{k_\ell } = k} $. Assuming that n, k1, …, kcare fixed, we ask: what is the minimum rate of the common message that the base-station needs to send so that the correct label is received at each of the k active users? This paper shows that instead of a conventional scheme of listing the indices of the users followed by their labels, which requires a common message rate of $k\left( {\log (n) + H\left( {\frac{{{k_1}}}{k}, \ldots ,\frac{{{k_c}}}{k}} \right)} \right)$ bits, it is possible to construct a fixed-length common message code with a rate of just $kH\left( {\frac{{{k_1}}}{k}, \ldots ,\frac{{{k_c}}}{k}} \right)$ bits plus a term that scales in n as O(log log(n)) for fixed k1, …, kc, where H(•) is the entropy of a probability distribution. If a variable-length code is permitted, the minimum common message rate is characterized as $kH\left( {\frac{{{k_1}}}{k}, \ldots ,\frac{{{k_c}}}{k}} \right) + O(1)$ bits, with no dependence on n. Finally, if k1, …, kcdeviate from the values for which the common message is designed, an additional cost per user equal to a Kullback-Leibler divergence term would be incurred.

This paper proposes a deep learning approach to channel sensing and downlink hybrid beamforming for massive multiple-input multiple-output systems operating in the time division duplex mode and employing either single-carrier or multicarrier transmission. The conventional precoding design involves a two-step process of first estimating the high-dimensional channel, then designing the precoders based on such estimate. This two-step process is, however, not necessarily optimal. This paper shows that by using a learning approach to design the analog sensing and the hybrid downlink precoders directly from the received pilots without the intermediate high-dimensional channel estimation, the overall system performance can be significantly improved. Training a neural network to design the analog and digital precoders simultaneously is, however, difficult. Further, such an approach is not generalizable to systems with different number of users. In this paper, we develop a simplified and generalizable approach that learns the uplink sensing matrix and downlink analog precoder using a deep neural network that decomposes on a per-user basis, then designs the digital precoder based on the estimated low-dimensional equivalent channel. Numerical comparisons show that the proposed methodology results in significantly less training overhead and leads to an architecture that generalizes to various system settings.

This paper shows that deep neural network (DNN) can be used for efficient and distributed channel estimation, quantization, feedback, and downlink multiuser precoding for a frequency-division duplex massive multiple-input multiple-output system in which a base station (BS) serves multiple mobile users, but with rate-limited feedback from the users to the BS. A key observation is that the multiuser channel estimation and feedback problem can be thought of as a distributed source coding problem. In contrast to the traditional approach where the channel state information (CSI) is estimated and quantized at each user independently, this paper shows that a joint design of pilots and a new DNN architecture, which maps the received pilots directly into feedback bits at the user side then maps the feedback bits from all the users directly into the precoding matrix at the BS, can significantly improve the overall performance. This paper further proposes robust design strategies with respect to channel parameters and also a generalizable DNN architecture for varying number of users and number of feedback bits. Numerical results show that the DNN-based approach with short pilot sequences and very limited feedback overhead can already approach the performance of conventional linear precoding schemes with full CSI.

Balancing traffic among network installed radio base stations is one of the main challenges facing mobile operators because of the unhomogeneous geographical distribution of mobile subscribers in addition to practical and environmental limitations preventing acquiring the best locations to build radio sites. This increases the challenge of satisfying the increasing data speed demand for smartphone users. In this paper, we present a reinforcement learning framework for optimizing neighbor cell relational parameters that can better balance the traffic between different cells within a defined geographical cluster. We present a comprehensive design of the learning framework that includes key system performance indicators and the design of a general reward function. System level simulations show that reinforcement learning based optimization for neighbor cell borders can significantly improve overall system performance; in particular, with a reward function defined as throughput, an improvement up to 50% is achieved.

In this paper, we consider the design of multi-resolution non-coherent multiple-input multiple-output (MIMO) systems that enable Unequal Error Protection (UEP). A method for designing multi-layer non-coherent Grassmannian constellations is introduced. Specifically, the proposed method yields multi-layer constellations that are amenable to a natural set partitioning strategy. The resulting subsets from such partitioning are used to encode the more protected symbols. On the other hand, the less protected symbols are mapped to points within these subsets. Furthermore, we present two methods to establish the link between the gain of the more protected layer and the design parameters of this construction. Finally, we exploit the underlying structure to develop a sequential decoding approach. Numerical results suggest that employing such decoding scheme leads to computational savings with respect to the optimal decoding while maintaining comparable performance.

In this paper, we develop a geometry-inspired methodology for generating systematic and structured Grassmannian constellations with large cardinalities. In the proposed methodology we begin with a small close-to-optimal “parent” Grassmann constellation. Each point in this constellation is augmented with a number of “children” points, which are generated along a set of geodesics emanating from that point. These geodesics are chosen to ensure close-to-maximal spacing. In particular, the directions of the geodesics and the distance that each “children” point is moved are chosen to maximize the pairwise Frobenius distance between the resulting constellation points. Although finding these directions directly seems difficult, by embedding the Grassmann manifold on a sphere of larger dimension, we were able to develop structures that are not only simple to generate but that also yield constellations that, under certain conditions, satisfy the maximum distance criterion and lie within a decaying gap from a tight upper bound. Numerical results suggest that the performance of the new constellations is comparable to that of the ones generated directly and significantly better than the performance of the ones generated using the exponential map.