On Multi-Step Sensor Scheduling via Convex Optimization

TL;DR

This paper introduces two efficient methods for multi-step sensor scheduling in linear Gaussian systems, leveraging convex optimization to approximate or find optimal schedules over long horizons.

Contribution

It proposes novel convex optimization-based approaches for sensor scheduling that reduce computational complexity compared to traditional binary integer programming.

Findings

The first method produces near-optimal sensor schedules efficiently.

The second method combines convex optimization with branch-and-bound for optimality.

Both methods outperform naive approaches in computational speed.

Abstract

Effective sensor scheduling requires the consideration of long-term effects and thus optimization over long time horizons. Determining the optimal sensor schedule, however, is equivalent to solving a binary integer program, which is computationally demanding for long time horizons and many sensors. For linear Gaussian systems, two efficient multi-step sensor scheduling approaches are proposed in this paper. The first approach determines approximate but close to optimal sensor schedules via convex optimization. The second approach combines convex optimization with a \BB search for efficiently determining the optimal sensor schedule.