Cross-Domain Continual Learning for Edge Intelligence in Wireless ISAC Networks

TL;DR

This paper tackles cross-domain sensing with CSI in edge-enabled ISAC networks under strict memory constraints. It introduces EdgeCL, a cross-domain continual learning framework that learns from sequential domain datasets using a transformer-based CSI sequence discriminator, distilled core-set replay for knowledge retention, and robustness-enhanced optimization to mitigate parameter drift. The approach achieves 89% of the cumulative-training performance while using only 3% of the memory and substantially reduces forgetting, demonstrating practical cross-domain sensing on HAR tasks in real ISAC deployments. This work enables scalable, memory-efficient edge intelligence for ubiquitous, adaptive sensing in 6G-era wireless networks.

Abstract

In wireless networks with integrated sensing and communications (ISAC), edge intelligence (EI) is expected to be developed at edge devices (ED) for sensing user activities based on channel state information (CSI). However, due to the CSI being highly specific to users' characteristics, the CSI-activity relationship is notoriously domain dependent, essentially demanding EI to learn sufficient datasets from various domains in order to gain cross-domain sensing capability. This poses a crucial challenge owing to the EDs' limited resources, for which storing datasets across all domains will be a significant burden. In this paper, we propose the EdgeCL framework, enabling the EI to continually learn-then-discard each incoming dataset, while remaining resilient to catastrophic forgetting. We design a transformer-based discriminator for handling sequences of noisy and nonequispaced CSI samples. Besides, we propose a distilled core-set based knowledge retention method with robustness-enhanced optimization to train the discriminator, preserving its performance for previous domains while preventing future forgetting. Experimental evaluations show that EdgeCL achieves 89% of performance compared to cumulative training while consuming only 3% of its memory, mitigating forgetting by 79%.

Figures (8)

• Figure 1: Working principle of EdgeCL: The CSI dataset of current domain and the knowledge learned from previous domains are jointly used for the cross-domain continual learning of EI.
• Figure 2: cap_new_fig2Diagram of the proposed algorithm, including the transformer-based discriminator, the robustness-enhanced optimization, and the distilled core-set selection.
• Figure 3: (a) Layouts of the experimental environments in a meeting room and (b) a lecture room.
• Figure 4: Comparison between the proposed algorithm and the benchmarks and baselines for the accuracy on each of the eight domain datasets after each training period, averaged over 30 trials.
• Figure 5: Impact of (a) number of domains $K$ and (b) number of exemplars per class on the average accuracy across the eight domains and 30 trials.