In this thesis, we consider a distributed detection problem in wireless sensor networks (WSNs) using ultrawide bandwidth (UWB) communications. Due to the severe restrictions on power consumption, energy efficiency becomes a critical design issue in WSNs. UWB technology has low-power transceivers, low-complexity and low-cost circuitry which are well suited to the physical layer requirements for WSNs. In a typical parallel fusion network, local decisions are made by local sensors and transmitted through a wireless channel to a fusion center, where the final decision is made. In this thesis, we control the number of UWB pulses to achieve the energy efficient distributed detection. We first theoretically characterize the performance of distributed detection using UWB communications. Both AWGN and fading channels are considered. Based on the analysis, we then obtain the minimum number of the pulses per detection to meet the required performance. To achieve a near-optimal design, we further propose a multiple access technique based on the random number of UWB pulses. Finally, the performance evaluation is provided to demonstrate the advantage of our design.

In this thesis, we consider decentralized multiclass classification problem in wireless sensor networks. In literature, the decentralized detection using error correcting code has been shown to have good fault-tolerance capability. In this thesis, we provide fault-tolerance capability by employing the code with a particular structure so that the decoding at the fusion center can be efficient. Specifically, the convolution code is employed to decode the local decision vector sent from all the local sensors. In addition, we proposed an efficient convolution code design algorithm by using simulated annealing. The simulation result shows that the proposed approach has good performance.

In this thesis, we consider the distributed classification problem in wireless sensor networks (WSNs). Sensor nodes in WSNs detect environmental variations and make their decisions individually, after which their decisions, possibly in the presence of faults, are transmitted to a fusion center. In literature, the distributed classification fusion using error correcting codes has been shown to have good sensor fault-tolerance capability. In this thesis, we extend the fault-tolerant classification system using error correcting code by using two-dimensional channel coding. We also extend the binary coding in literature to the M-ary code. This thesis then suggests a code construction method with low computational complexity. Based on the suggest code construction method, this thesis then conducts a series experiment to investigate the performance of the suggested method.

We introduce the notion of diversity order as a performance measure for distributed radar systems. We define the diversity order of a radar network as the slope of the probability of detection (PD) versus SNR evaluated at PD =0.5. We prove that the communication bandwidth between the sensors and the fusion center does not affect the growth in diversity order. We also prove that the OR rule leads to the best performance and its diversity order grows as (log K). We then introduce the notion of a random radar network to study the effect of geometry on overall system performance. We approximate the distribution of the SINR at each sensor by an exponential distribution, and we derive the moments for a specific system model. We then analyze multistatic systems and prove that each sensor should be large enough to cancel the interference in order to exploit the available spatial diversity.

We introduce the notion of diversity order as a performance measure for distributed radar systems. We define the diversity order of a radar network as the slope of the probability of detection (PD) versus SNR evaluated at PD =0.5. We prove that the communication bandwidth between the sensors and the fusion center does not affect the growth in diversity order. We also prove that the OR rule leads to the best performance and its diversity order grows as (log K). We then introduce the notion of a random radar network to study the effect of geometry on overall system performance. We approximate the distribution of the SINR at each sensor by an exponential distribution, and we derive the moments for a specific system model. We then analyze multistatic systems and prove that each sensor should be large enough to cancel the interference in order to exploit the available spatial diversity.

This work considers the sampling design for detection problems.Firstly,we focus on studying the effect of signal shape on sampling design for Gaussian detection problem.We then investigate the sampling design for distributed detection problems and compare the performance with the single sensor context. We also propose a sampling design scheme for the cluster-based wireless sensor networks.The cluster head employs a linear combination fusion to reduce the dimension of the sampled observation.Mathematical verification and simulation result show that the performance loss caused by the dimensionality reduction is exceedingly small as compared with the benchmark scheme,which is the sampling scheme without dimensionality reduction.In particular,there is no performance loss when the identical sampling points are employed at all sensor nodes.

The energy efficiency issue, which is subjected to an energy constraint, is important for the applications in wireless sensor network. For the distributed detection problem considered in this thesis, the sensor makes a local decision based on its observation and transmits a one-bit message to the fusion center. We consider the local sensors employing a censoring scheme, where the sensors are silent and transmit nothing to fusion center if their observations are not very informative. The goal of this thesis is to achieve an energy efficiency design when the distributed detection employs the censoring scheme. Simulation results show that we can have the same error probabilities of decision fusion while conserving more energy simultaneously as compared with the detection without using censoring schemes. In this thesis, we also demonstrate that the error probability of decision fusion is a convex function of the censoring probability.

Recently, distributed detection has been intensively studied. The prevailing model for distributed detection (DD) is a system involving both distributed local sensors and a fusion center. In a DD system, multiple sensors work collaboratively to distinguish between two or more hypotheses, e.g., the presence or absence of a target. In this thesis, the classical DD problem is reexamined in the context of wireless sensor network applications. For minimize the error probability at the fusion center, we consider the conventional method that designs the optimal binary local sensor decision rule in a channel-aware system, i.e., it integrates the transmission channel characteristics for find the optimal binary local sensor decision threshold to minimize the error probability at the fusion center. And there have different optimal local sensor decision thresholds for different channel state information. Because of optimal multi-bit (soft) local sensor decision is more practical than optimal binary local sensor decision. Allowing for multi-bit local sensor output, we also consider another conventional method that designs the optimal multi-bit (soft) local sensor decision rule in a channel-aware system. However, to design the optimal local sensor decision rule, both of two conventional methods are easily trapped into local optimal thresholds, which are depended on the pre-selected initialization values. To overcome this difficulty, we consider several modified Simulated Annealing (SA) algorithms. Based on these modified SA algorithms and two conventional methods, we propose two novel SA algorithms for implementing the optimal local sensor decision rule. Computer simulation results show that the employments of two novel SA algorithms can avoid trapping into local optimal thresholds in both optimal binary local sensor decision problem and optimal multi-bit local sensor decision problem. And two novel SA algorithms offer superior performance with lower search points compared to conventional SA algorithm.

We design, analyze, and optimize distributed detection and estimation algorithms in a large, shared-channel, single-hop wireless sensor network (WSN). The fusion center (FC) is allocated a shared transmission channel to collect local decisions/estimates but cannot collect all of them because of limited energy, bandwidth, or time. We propose a strategy called reliability-based splitting algorithm that enables the FC to collect local decisions/estimates in descending order of their reliabilities through a shared collision channel. The algorithm divides the transmission channel into time frames and the sensor nodes into groups based on their observation reliabilities. Only nodes with a specified range of reliabilities compete for the channel using slotted ALOHA within each frame. Nodes with the most reliable decisions/estimates attempt transmission in the first frame; nodes with the next most reliable set of decisions/estimates attempt in the next frame; etc. The reliability-based splitting algorithm is applied in three scenarios: time-constrained distributed detection; sequential distributed detection; and time-constrained estimation. Performance measures of interest - including detection error probability, efficacy, asymptotic relative efficiency, and estimator variance - are derived. In addition, we propose and analyze algorithms that exploit information from the occurrence of collisions to improve the performance of both time-constrained distributed detection and sequential distributed detection.

The introduction of programmability and dynamic protocol deployment in routers, there would be an increase in the potential vulnerabilities and attacks . The next- generation Internet promises to provide a fundamental shift in the underlying architecture to support dynamic deployment of network protocols. In this thesis, we consider the problem of detecting malicious packet drops in routers. Specifically, we focus on an attack scenario, where a router selectively drops packets destined for another node. Detecting such an attack is challenging since it requires differentiating malicious packet drops from congestion-based packet losses. We propose a controller- based malicious packet detection technique that effectively detects malicious routers using delayed sampling technique and verification of the evidence. The verification involves periodically determining congestion losses in the network and comparing the forwarding behaviors of the adjoining routers to affirm the state of a router in the network. We provide a performance analysis of the detection accuracy and quantify the communication overhead of our system. Our results show that our technique provides accurate detection with low performance overhead.