PIB: Prioritized Information Bottleneck Framework for Collaborative Edge Video Analytics
Zhengru Fang, Senkang Hu, Liyan Yang, Yiqin Deng, Xianhao Chen, Yuguang Fang
TL;DR
The paper tackles efficient collaborative edge video analytics under limited bandwidth by proposing a Prioritized Information Bottleneck (PIB) framework. PIB selects task-relevant features by a priority mechanism that accounts for channel quality via $C_k = B_k \log_2(1 + \text{SNR}_k)$ and region-of-interest coverage, then compresses data through a deterministic information bottleneck with variational bounds. A multi-camera weighted IB objective with $I_w$ terms and a multi-frame correlation model captures temporal redundancy and enables end-to-end training using variational approximations of $I(Z^{(k)};Y^{(k)})$ and $I(X^{(k)};Z^{(k)})$. Empirical results on simulated urban pedestrian scenarios show PIB achieves up to 15.1% MODA improvement and 66.7% lower communication cost compared to TOCOM-TEM, JPEG, and HEVC, while maintaining low latency. The approach enables scalable, edge-based collaborative perception without reconstructing videos, offering practical benefits for autonomous driving and real-time edge analytics.
Abstract
Collaborative edge sensing systems, particularly in collaborative perception systems in autonomous driving, can significantly enhance tracking accuracy and reduce blind spots with multi-view sensing capabilities. However, their limited channel capacity and the redundancy in sensory data pose significant challenges, affecting the performance of collaborative inference tasks. To tackle these issues, we introduce a Prioritized Information Bottleneck (PIB) framework for collaborative edge video analytics. We first propose a priority-based inference mechanism that jointly considers the signal-to-noise ratio (SNR) and the camera's coverage area of the region of interest (RoI). To enable efficient inference, PIB reduces video redundancy in both spatial and temporal domains and transmits only the essential information for the downstream inference tasks. This eliminates the need to reconstruct videos on the edge server while maintaining low latency. Specifically, it derives compact, task-relevant features by employing the deterministic information bottleneck (IB) method, which strikes a balance between feature informativeness and communication costs. Given the computational challenges caused by IB-based objectives with high-dimensional data, we resort to variational approximations for feasible optimization. Compared to TOCOM-TEM, JPEG, and HEVC, PIB achieves an improvement of up to 15.1\% in mean object detection accuracy (MODA) and reduces communication costs by 66.7% when edge cameras experience poor channel conditions.