Vimal Radhakrishnan

Partial distributed beamforming design in passive radar sensor networks

In this papers we address the optimal power allocation problem in a distributed passive radar sensor network system, where closely located nodes are capable of distributed beamforming. In this setup, the network of sensor nodes is viewed as a combination of node clusters, where each cluster is capable of time synchronization, and coordinated transmission. The goal of the network is to provide a reliable estimation of a source signal, by collecting and combining the individual observations from the sensor nodes in a centralized node. In this regard, a minimum mean squared error (MMSE) problem is formulated for the class of unbiased estimators. As it is shown, the resulting problem can be formulated as a convex optimization problem, which is solvable with the standard numerical solvers. At the end, numerical simulations illustrate the effect of the different network parameters on the resulting performance, and a significant gain is observed by enabling the proposed distributed beamforming algorithm for multiple sensor clusters.

Widely-linear processing for distributed passive radar systems with strictly non-circular sources.

In this paper, we address the design of a distributed passive radar system which operates on a strictly non-circular source. The goal of a passive distributed radar is to provide a reliable estimation from the source signal, by collecting and combining the individual observations from the network in a centralized node. We take advantage of the non-circular nature of the source, and propose widely linear signal processing algorithms, which achieve around 3 dB of gain in the estimation accuracy compared to the available linear strategies. In this regard, a minimum mean squared error (MMSE) problem is formulated for unbiased class of estimators, where the widely-linear processing is enabled at the distributed sensors, or at a centralized entity. An optimal power allocation and information fusion is achieved in each case, by studying the optimality conditions of the corresponding problems. The performance of the proposed methods are then compared to the known linear strategies via numerical simulations.

Differential Multidimensional Scaling for Self-Localization of TDOA Sensor Networks

We present a novel algorithm for self-localization in sensor networks without any prior knowledge on the locations of the sensors. We assume that all sensors in the network can receive and transmit, thus we obtain time difference of arrival measurements for all combinations of sensors. Using the full set of these differences in arrival times in the network we are able to obtain the relative location of the sensors nodes, the shape of the network. This leaves us with the problem of anchoring the network to its absolute location, which we solve using additional transmitting beacons at known locations. Experimental results from numerical simulation demonstrate the performance of our approach under various conditions.

Optimal linear MMSE design for passive distributed radar sensor network systems

In this paper we present optimal power allocation, together with optimal linear signal fusion, considering a passive distributed radar sensor network system. The goal of a passive distributed radar is to obtain a reliable estimation from a source signal, by collecting and combining the individual observations from the network of sensor nodes (SN)s in a fusion center (FC). In this respect, a linear minimum-mean-square-error (MMSE) optimization strategy is considered, where optimal linear operation at the SNs as well as the FC are obtained analytically. The obtained solutions are then analytically and numerically compared to the previously studied unbiased linear MMSE (ULMMSE) approach. It is observed that both schemes share the same strategy for the optimal power allocation among the SNs, but differ in the corresponding linear fusion. As expected, the proposed approach reaches a lower estimation MSE compared to the ULMMSE one.

Subspace-based adaptive widely linear blind channel estimation for constrained minimum variance CDMA receiver

We propose a subspace-based Widely Linear (WL) blind channel estimation scheme based on the iterative power method for the WL constrained minimum variance Code Division Multiple Access (CDMA) receiver. The novel technique approximates the noise subspace by using a matrix power and the WL processing fully exploits the second-order non-circularity of the signal. Two adaptive recursive least squares algorithms are developed using power iterations, which completely avoid the computationally intensive singular value decomposition. Simulation results show an improved performance of the proposed algorithms in terms of convergence and complexity as compared to their linear counterparts.

Distributed beamforming for cooperative networks with widely-linear processing at the relays and the receiver

This paper addresses the distributed beamforming problem, where widely-linear (WL) processing is employed at both the relays and the receiver to take advantage of strictly second-order (SO) noncircular source signals. We consider a single-antenna communication pair in a relay network, which suffers from strong interference. Assuming perfect channel state information (CSI), we design two algorithms based on the maximization of the signal-to-interference-plus-noise ratio (SINR) under a total relay power constraint. While the first algorithm jointly optimizes the weights at the relays and the receiver using semidefinite relaxation (SDR), the second algorithm performs a separate optimization in closed-form, requiring a substantially lower cost, but yielding almost the same performance. We show through simulations that the respective performance improvements associated with the WL processing at the relays and the receiver accumulate such that significant gains can be achieved compared to linear processing. Also, the complexity of the two algorithms is analyzed.