Optimal Sensor Collaboration for Parameter Tracking Using Energy Harvesting Sensors

Abstract

Recent advances in wireless communications and electronics have enabled the development of low-cost, low-power, multifunctional sensor nodes that are deployed as networks for many applications such as environment monitoring, source localization and target tracking [1]- [3]. These sensor nodes sense and measure some attributes of the targets of interest, and are able to exchange their information through innetwork communications [4]. In this paper, we study the problem of parameter tracking in the presence of inter-sensor communications that is referred to as sensor collaboration.Here sensors are equipped with energy harvesters, which can replenish energy from the environment, such as sun light and wind. The act of sensor collaboration allows sensors to update

Joint 2D‐DOA and polarization estimation for Electromagnetic vector sensors array with compressive measurements

“…As the compression matrix is sparse according to the assumptions given in Section 2, the optimization problem involving Φ ∈ C N ×6 M turns out to be a non‐convex problem which is generally intractable (NP‐Hard). To handle this problem, motivated by [29, 30], we build up the relationship between the sparse compression matrix topology and the compressive coefficients. Without loss of generality, we use the notation $$\mathbf{W}\in {\mathbb{C}}^{P\times Q}$$ to denote the compression matrix.…”

Joint Collaboration and Compression Design for Distributed Sequential Estimation in a Wireless Sensor Network

“…the collaboration weight at time k, ⊗ denotes the Kronecker product and I L is the L-dimensional identity matrix. Meanwhile, to further reduce the communication cost between the local sensors and the FC, the observations from sensor nodes with i ∈ [1, M ] are linearly compressed [11], [17] as follows:…”

“…As the structure of W(k) may be sparse which is determined by the sensor network topology, it makes the problem (26) hard to solve. To overcome this problem, motivated by [17], we establish the correspondence between the structure of the sparse network topology and collaboration weights. Specifically, we first vectorize the collaboration matrix and eliminate all the elements whose corresponding entry in the network topology matrix A equals to zero, then constitute a new vector w ∈ R U×1 where U is the total number of nonzero elements in A.…”

“…The optimal sensor collaboration strategy is designed based on the prior knowledge about parameter correlations. Recently, the work in [17] studied the tracking of a dynamic parameter which follows a first-order Gauss-Markov process with an energyconstrained sensor network. The optimal sensor collaboration strategy is designed in the presence of noisy sensor to sensor communication channels.…”

“…Noisy sensor collaboration Compression In this section, we describe the problem of parameter estimation based on the collaboration-compression scheme. In the proposed framework, instead of transmitting all the observations from individual sensor nodes to the FC directly, each sensor node first performs spatial collaboration via a coherent MAC [13], [17] after observing the parameter of interest through a linear measurement model. Then the measurements after collaboration are linearly compressed [11] and transmitted to the FC through a MAC.…”