Cross-Layer Design of Distributed Sensing-Estimation With Quality Feedback— Part I: Optimal Schemes

Abstract

This two-part paper presents a feedback-based crosslayer 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 first part, the optimization complexity is reduced by exploiting the statistical symmetry and large network approximation of the WSN. Structural properties of the optimal policy are derived for a coordinated and a decentralized scheme. It is proved that a dense WSN provides sensing diversity, so that only a few SNs with the best local observation quality need to be activated, despite the fluctuations of the WSN. The optimal policy dictates that, when the estimation quality is poor, only the best SNs activate, otherwise all SNs remain idle to preserve energy. The costs of coordination and feedback are evaluated, revealing the scalability of the decentralized scheme to large WSNs, at the cost of performance degradation. Simulation results demonstrate cost savings from 30% to 70% over a nonadaptive scheme, and significant gains over a previously proposed estimator which does not consider these cross-layer factors.

Distributed remote estimation over the collision channel with and without local communication

“…In [14], [15], a Bayesian framework is developed for decentralized remote sensing, which adapts the quantization rate of transmissions based on feedback from fusion center. In [16], [17], the fusion center feeds back estimation quality that is coupled with local sensor quality to jointly optimize cross-layer performance. Combined with reinforcement learning, several works designed optimal strategies for more realistic systems with feedback [21]- [23].…”

Optimal Adaptive Random Multiaccess in Energy Harvesting Wireless Sensor Networks

“…where x [i] (e, λ) is given by (33). The solution (32) directly follows by exploiting the strict concavity of g(x) and the fact that g (x) = V (y th (x)).…”

Cross-Layer Estimation and Control for Cognitive Radio: Exploiting Sparse Network Dynamics

“…. Assuming a collision model for the B control channels and N S → ∞, the number of measurements received at the CC, M k , has binomial distribution with B trials and success probability ψ k e −ψ k [20], i.e.,…”

“…Active sensor scheduling and adaptation [15] encompass applications such as target tracking [16], [17], physical activity detection [18], and sequential hypothesis testing [19]. All these prior works including ours [20]- [22] assume that the underlying state is given by nature and is not controlled. In contrast, in this work, states are affected by scheduling decisions, via interference and collisions generated by the SUs to the PUs, and we design joint controlled sensing, estimation and scheduling schemes in wireless networks, which account for the cost of acquisition of state information and its impact on the overall network performance.…”