While distributed multi-cell resource allocation (D-MCRA) is promising for improving the spectral efficiency of cellular systems, it is challenging to realize in practice due to the large overhead for exchanging the channel state information (CSI) between base stations (BSs) and limited backhaul bandwidth. Inspired by the emerging integrated sensing and communication (ISAC) technique for mmWave systems, we propose in this paper a new distributed user scheduling and beamforming framework with a small signaling overhead. Specifically, we employ the ISAC signal for the BSs to track the kinematic parameters of the served users, which not only can be used to estimate the line-of-sight (LoS) CSI of served users but also can be exchanged with other BSs to construct cross-cell CSI. Based on this, we propose an enhanced proportional fairness zero-forcing greedy (PFZFG) scheduler for BSs to determine the set of users for ISAC signal transmission distributively. Afterward, each BS optimizes the ISAC beamformers based on a signal-to-average-leakage-plus-interference-plus-noise ratio (SALINR) criterion and subject to sensing error constraints. Simulation results show that our proposed design can approach the performance of the centralized algorithm.

Acquiring accurate channel state information (CSI) is challenging in dynamic millimeter wave networks due to the excessive signaling overhead. In this work, we leverage the integrated sensing and communication (ISAC) technique for simultaneous data communication and CSI acquisition via proactive sensing. While offering a low overhead solution, ISAC beamforming design faces a complex trade-off between sensing accuracy and communication interference management. To handle it, we first formulate the ISAC beamforming design as an optimization problem that aims to minimize the worst sensing error subject to robust transmission rate constraints in the presence of CSI acquisition error. We then propose a benchmark algorithm by successive convex approximation. Further, we develop a low-complexity algorithm via dual optimization and max-min optimization methods. Simulation results demonstrate that the proposed design greatly outperforms existing schemes and achieves robust transmission in dynamic scenarios.

Incremental learning could be roughly divided into two categories, i.e., class- and task-incremental learning. The main difference is whether the task ID is given during evaluation. In this paper, we show this task information is indeed a strong prior knowledge, which will bring significant improvement over class-incremental learning baseline, e.g., DER [39]. Based on this observation, we propose a gate network to predict the task ID for class incremental inference. This is challenging as there is no explicit semantic relationship between categories in the concept of task. Therefore, we propose a multi-centroid task descriptor by assuming the data within a task can form multiple clusters. The cluster centers are optimized by pulling relevant sample-centroid pairs while pushing others away, which ensures that there is at least one centroid close to a given sample. To select relevant pairs, we use class prototypes as proxies and solve a bipartite matching problem, making the task descriptor representative yet not degenerate to uni-modal. As a result, our dynamic inference network is trained independently of baseline and provides a flexible, efficient solution to distinguish between tasks. Extensive experiments show our approach achieves state-of-the-art results, e.g., we achieve 72.41% average accuracy on CIFAR100-BOS50, outperforming DER by 3.40%.

Decentralized multi-cell coordinated beamforming (D-MCBF) is a promising technique to improve the system spectral efficiency, but it usually requires the base stations (BSs) to frequently exchange large amounts of information in order to approach the centralized MCBF solution. However, in practical environments with time-varying channels and limited backhaul bandwidth, frequent information exchange causes delays and thus the existing D-MCBF methods suffer significant performance loss. In this paper, we aim to design a D-MCBF method that can approach the centralized MCBF solution while with only a few inter-BS information exchanges. By assuming that the BSs can exchange the interference channel powers, we firstly formulate a virtual power control based sum rate maximization (VPC-SRM) problem where each BS individually optimizes the beamformers for its served users and at the same time “virtually” optimizes the transmission powers of other BSs. Thus, the VPC-SRM problem is a surrogate of the centralized SRM problem and can be efficiently handled by existing iterative algorithms. To provide a good initial point, we further propose a fully decentralized leakage-based SRM formulation. Simulation results show that the proposed D-MCBF algorithms can perform closely with the centralized method with only two times of information exchange even when the channel is time-varying.