Beam Alignment in Multipath Environments for Integrated Sensing and Communication using Bandit Learning

TL;DR

This paper tackles rapid beam alignment for millimeter-wave communications by integrating sensing and communication (ISAC) with multi-armed bandit (MAB) learning. The core idea is to use radar information to prune the candidate beam set from $K$ to a smaller $\tilde{K}$ and to forecast realignment needs in quasi-stationary scenarios, enabling the UCB algorithm to operate on a reduced search space. The authors implement radar signal processing and the MAB logic on a system-on-chip (SoC) platform with hardware-software co-design and fixed-point analysis, achieving a notable reduction in exploration time (about 35%) and a throughput improvement of approximately 1.4× over conventional MAB that relies solely on communications metrics. Through simulations and hardware timing results, the work demonstrates improved BER/throughput under realistic multipath and clutter, and provides a practical pathway to deploy ISAC-MAB in edge devices and many-beam mmWave systems, with potential extensions to KL-UCB, Thompson sampling, and adversarial bandits.

Abstract

Prior works have explored multi-armed bandit (MAB) algorithms for the selection of optimal beams for millimeter-wave (mmW) communications between base station and mobile users. However, when the number of beams is large, the existing MAB algorithms are characterized by long exploration times, resulting in poor overall communication throughput. In this work, we propose augmenting the upper confidence bound (UCB) based MAB with integrated sensing and communication (ISAC) to address this limitation. The premise of the work is that the radar and communication functionalities share the same field-of-view and that communication mobile users are detected by the radar as mobile targets. The radar information is used for significantly reducing the number of candidate beams for the UCB, resulting in an overall reduction in the exploration time. Further, the radar information is used to estimate the realignment time in quasi-stationary scenarios. We have realized the MAB and radar signal processing algorithms on the system on chip (SoC) via hardware-software co-design (HSCD) and fixed-point analysis. We demonstrate the significant gain in execution time using accelerators. The simulations consider complex propagation channels involving direct and multipath, with simple and extended radar targets in the presence of significant static clutter. The resulting experiments show that the proposed ISAC-based MAB achieves a 35% reduction in the overall exploration time and 1.4 factor higher throughput as compared to the conventional MAB that is based only on communications.

Figures (10)

• Figure 1: Duty cycle of radar search time over total time due to shared functionality with communications.
• Figure 2: IEEE 802.11ad PHY frame with radar waveform embedded in it.
• Figure 3: Block diagram of ISAC-MAB algorithm.
• Figure 4: Correlation output of $\mathop{\mathrm{\mathbf{s}_{tx}}}\nolimits$ with a perfectly matched radar signal, noise, and the mismatched preamble of the uplink communication signal.
• Figure 5: (a) Timing diagram for beam selection for communications (COMM) based on UCB algorithm (b).i. Timing diagram for beam selection for proposed ISAC framework using radar waveform (RW), radar signal processing (RSP) and UCB, (b).ii. Radar waveforms with 50% duty cycle for multiple pulse repetition interval $T_p$ over coherent processing interval (CPI).