Learning and Communications Co-Design for Remote Inference Systems: Feature Length Selection and Transmission Scheduling

TL;DR

This paper presents a joint learning and communication co-design framework for remote inference systems, optimizing feature length and transmission scheduling to significantly reduce inference error.

Contribution

It introduces a novel co-design approach for feature length selection and transmission scheduling in remote inference, including optimal solutions for single-channel scenarios and a low-complexity algorithm for multi-channel cases.

Findings

Optimal co-designs reduce inference error significantly.

Proposed algorithms outperform baseline methods.

Trace-driven evaluations show up to 10,000 times error reduction.

Abstract

In this paper, we consider a remote inference system, where a neural network is used to infer a time-varying target (e.g., robot movement), based on features (e.g., video clips) that are progressively received from a sensing node (e.g., a camera). Each feature is a temporal sequence of sensory data. The inference error is determined by (i) the timeliness and (ii) the sequence length of the feature, where we use Age of Information (AoI) as a metric for timeliness. While a longer feature can typically provide better inference performance, it often requires more channel resources for sending the feature. To minimize the time-averaged inference error, we study a learning and communication co-design problem that jointly optimizes feature length selection and transmission scheduling. When there is a single sensor-predictor pair and a single channel, we develop low-complexity optimal co-designs for both the cases of time-invariant and time-variant feature length. When there are multiple sensor-predictor pairs and multiple channels, the co-design problem becomes a restless multi-arm multi-action bandit problem that is PSPACE-hard. For this setting, we design a low-complexity algorithm to solve the problem. Trace-driven evaluations demonstrate the potential of these co-designs to reduce inference error by up to 10000 times.