Cross-layer design of distributed sensing-estimation with quality feedback, Part II: Myopic schemes

TL;DR

This paper develops low-complexity myopic sensing-transmission policies for wireless sensor networks to optimize estimation performance and energy use, considering feedback and cross-layer factors, with proven near-optimal results.

Contribution

It introduces closed-form and iterative algorithms for myopic policies in both coordinated and decentralized schemes, improving practicality over previous optimal policies.

Findings

Thresholds for sensor activation are derived in closed form.

Single-channel operation suffices for energy-constrained networks.

Proposed policies achieve near-optimal performance with lower complexity.

Abstract

This two-part paper presents a feedback-based cross-layer framework for distributed sensing and estimation of a dynamic process by a wireless sensor network (WSN). Sensor nodes wirelessly communicate measurements to the fusion center (FC). Cross-layer factors such as packet collisions and the sensing-transmission costs are considered. Each SN adapts its sensing-transmission action based on its own local observation quality and the estimation quality feedback from the FC under cost constraints for each SN. In this second part, low-complexity myopic sensing-transmission policies (MPs) are designed to optimize a trade-off between performance and the cost incurred by each SN. The MP is computed in closed form for a coordinated scheme, whereas an iterative algorithm is presented for a decentralized one, which converges to a local optimum. The MP dictates that, when the estimation quality is poor, only the best SNs activate, otherwise all SNs remain idle to preserve energy. For both schemes, the threshold on the estimation quality below which the SNs remain idle is derived in closed form, and is shown to be independent of the number of channels. It is also proved that a single channel suffices for severely energy constrained WSNs. The proposed MPs are shown to yield near-optimal performance with respect to the optimal policy of Part I, at a fraction of the complexity, thus being more suitable for practical WSN deployments.