Wireless Sensor Networks are well suited for tracking targets carrying RFID tags in indoor environments. Tracking based on the received signal strength indication (RSSI) is by far the cheapest and simplest option, but suffers from secular biases due to effects of multi-path, occlusions and decalibration, as well as large unbiased errors due to measurement noise. We propose a novel algorithm that solves these problems in a distributed, scalable and power-efficient manner. Firstly, our proposal includes a tandem incremental estimator that learns and tracks the radio environment of the network, and provides this knowledge for the use of the tracking algorithm, which eliminates the secular biases due to radio occlusions etc. Secondly, we reduce the unbiased tracking error by exploiting the co-dependencies in the motion of several targets (as in crowds or herds) via a fully distributed and tractable particle filter. We thereby extract a significant 'diversity gain' while still allowing the network to scale seamlessly to a large tracking area. In particular, we avoid the pitfalls of network congestion and severely shortened battery lifetimes that plague procedures based on the joint multi-target probability density.

Free-space optical (FSO) transmission systems enable high-speed communication with relatively small deployment costs. However, FSO suffers a critical disadvantage, namely susceptibility to fog, smoke, and conditions alike. A possible solution to this dilemma is the use of hybrid systems employing FSO and radio frequency (RF) transmission. In this paper we propose the application of a rateless coded automatic repeatrequest scheme for such hybrid FSO/RF systems. The advantages of our approach are (a) the full utilization of available FSO and RF channel resources at any time, regardless of FSO or RF channel conditions and temporal variations, and (b) no need for a-priori rate selection at the transmitter. In order to substantiate these claims, we establish the pertinent capacity limits for hybrid FSO/RF transmission and present simulation results for transmission with off-the-shelf Raptor codes, which achieve realized rates close to these limits under a wide range of channel conditions. We also show that in conditions of strong atmospheric turbulence, rateless coding is advantageous over fixed-rate coding with rate adaptation at the transmitter.

We have recently proposed a novel receiver for Ultra-Wide-band Impulse-Radio communication in bursty applications like Wireless Sensor Networks. The receiver, based on the principle of Compressed Sensing (CS), exploits the sparsity of the transmitted signal to achieve reliable demodulation. It acquires a modest number of projections of the received signal using analog correlators, and performs a joint decoding of the time of arrival and the data bits from these under-sampled measurements via an efficient quadratic program. In this paper we examine the robustness of this receiver to strong narrow-band interference (NBI) from primary licensed systems like WiMAX. First, by choosing frequency selective test functions in the front-end correlators, we ensure that the interferer can corrupt only a small fraction of the CS measurements. Then we implement a 'digital notch' by identifying and dropping those affected measurements during the quadratic programming reconstruction. The method is easily extended to multiple interferers without additional cost or complexity. We show that by implementing such a 'digital notch' the receiver becomes extremely robust to NBIs. For example its performance is negligibly affected even when the WiMAX customer premise equipment is at a distance comparable to that of the UWB transmitter and the base station is only ten times farther off, both very practical scenarios.

We propose a novel technique for tracking multiple co-dependently maneuvering targets using a wireless sensor network. We consider the scenario where the targets carry radio frequency identification (RFID) tags and the sensors in the network measure some metric of the radio transmissions from these tags, like the received signal strength, the time of arrival or the angle of arrival. These measurements are then processed by a sampling importance re-sampling particle filter for tracking. While such a set-up is now fairly standard in literature, the novel aspect of our algorithm is that it exploits the co-dependencies in the motion of the targets via a fully distributed and tractable particle filter bank. We thereby extract a significant "diversity gain", while allowing the network to scale seamlessly to a large tracking region. In particular, we avoid the pitfalls of network congestion and severely shortened battery lifetimes that plague currently used procedures that implement the filter on the joint multi-target probability density.

A Wireless Sensor Network (WSN) observes a natural field and aims to recreate it with sufficient fidelity at a, perhaps distant, Fusion Center (FC) using a wireless communication channel of arbitrary capacity. We propose a universal and power efficient method for such data extraction, based on Digital Fountain Codes (DFCs) and joint-source channel decoding. Our method implements a distributed `rate-lessiquest DFC which automatically tunes the number of transmissions to the channel capacity. Furthermore, instead of directly compressing the WSN data, we achieve rate reduction by treating the spatiotemporal dependencies in the field as an outer code, and jointly decoding this concatenation at the FC using a multi-stage iterative decoder. We demonstrate that a power efficiency close to the capacity-rate-distortion limit is achieved at moderate distortion levels, irrespective of the channel capacity or field dependencies. As compared to the traditional approach of source-channel separation, the proposed data extraction scheme is particularly attractive for WSN applications due its computationally simple encoding procedure, low latency and the ability to seamlessly trade-off fidelity of reconstruction for power consumption.

Fountain codes have recently gained wide attention in communications due to their capacity-approaching performance and rateless properties that allow them to seamlessly adapt to unknown channel statistics. In this paper, we consider the problem of low complexity decoding of Luby transform codes, which are a class of linear fountain codes. We adapt the recently proposed technique of informed dynamic scheduling to the rateless regime, and combine it with the method of incremental decoding to obtain a decoder that has a significantly reduced computational load compared to the commonly used alternative of message-reset decoding with a flooding schedule. This reduction in complexity, in some cases as large as a factor of sixty, is obtained without affecting the error rate performance of the code.

We investigate an 'inference first' (IF) approach to information retrieval from a wireless sensor network (WSN). In this method, statistical estimation pertinent to the user's application is implemented within the network (in-situ) and only the relevant sufficient statistics are exported. We formulate this procedure as a delay-free filtering problem on a spatio-temporal hidden Markov model (HMM), and propose a scalable approximate distributed filter. The algorithm is a novel application of the idea of iterated decoding, where we iteratively marginalize the joint distribution of the state of the HMM at two consecutive time epochs. We compare and contrast algorithms like the Gibbs sampler (GS), mean field decoding (MFD) and broadcast belief propagation (BBP), and discuss their suitability for in-situ marginalization. A simplified analysis of the energy gain achievable by the IF approach, relative to centralized processing, is provided.

In many lifetime enhancement strategies for wireless sensor networks (WSNs) it is often necessary to identify the statistical model of the underlying physical field. We consider the problem of in-situ inference as an exemplary application and propose an in-situ model estimation algorithm that works in tandem with a parametric distributed filtering procedure. We demonstrate, via averaged-gradient analysis and simulations, that the resulting adaptive filter is stable, robust and, importantly, fully scalable. It compares favorably with kernel-regression inference, and typically significantly outperforms the latter when the spatio-temporal variations in the natural field are relatively rapid.