Wireless Sensor Network (WSN) has become a popular wireless technology in recent years. In WSN, a large number of sensors are used to collect data and forward data hop-by-hop to a sink. Due to the unbalancing of traffic load, some grid nodes may consume more energy and their packet loss ratio may be increased as well. In order to improve above-mentioned shortcomings, in this Thesis, we propose an Adaptive Grid-based Data Collection (AGDC) scheme. Because a mobile sink may move, it is possible the traffic load of primary grid nodes can be changed in WSN. According to the distribution of traffic load, the AGDC can adjust transmission range to allocate one or more temporary grid nodes between two primary grid nodes. Through the added temporary grid nodes, traffic load is evenly dispersed among different grid nodes. We allow the primary grid nodes to use smaller transmission power to save energy and allow the temporary grid nodes to buffer data to reduce packet loss ratio. For the purpose of evaluation, we perform simulation on NS-2. With the proposed AGDC scheme, the transmission range of a primary grid node can be set to an appropriate distance to reduce power consumption and packet loss ratio. Since the packet loss ratio is reduced, the throughput of entire WSN is increased.

A local reconfiguration algorithm (INP) for reliable routing in wireless sensor networks that consist of many static (fixed) energy-constrained nodes is introduced in the dissertation. For routing around crash fault nodes, a communication tree structure connecting sensor nodes to the base station (sink or root) is dynamically reconfigured during information dissemination. Unlike other location based routing approaches, INP does not take any support from a high costing system that gives position information such as GPS. For reconfigurations, INP uses only local relational information in the tree structure among nearby nodes by collaboration between the nodes that does not need global maintenance, so that INP is energy efficient and it scales to large sensor networks. The performance of the algorithm is compared to the single path with repair routing scheme (SWR) that uses a global metric and the modified GRAdient broadcast scheme (GRAB-F) that uses interleaving multiple paths by computation and by simulations. The comparisons demonstrate that using local relative information is mostly enough for reconfigurations, and it consumes less energy and mostly better delivery rates than other algorithms especially in dense environments. For the control observer to know the network health status, two new diagnosis algorithms (Repre and Local) that deal with crash faults for wireless sensor networks are also introduced in the dissertation. The control observer knows not only the static faults found by periodic testing but also the dynamic faults found by a path reconfiguration algorithm like INP that is invoked from evidence during information dissemination. With based on this information, the control observer properly treats the network without lateness. Local algorithm is introduced for providing scalability to reduce communication energy consumption when the network size grows. The performance of these algorithms is computationally compared with other crash faults identification algorithm (WSNDiag). The comparisons demonstrate that maintaining the communication tree with local reconfigurations in Repre and Local needs less energy than making a tree per each diagnosis procedure in WSNDiag. They also demonstrate that providing scalability in Local needs less energy than other approaches.

Much research has been done on wireless sensor networks. However, most protocols and algorithms for such networks are based on the ideal model Unit Disk Graph (UDG) model or do not assume any model. Furthermore, many results assume the knowledge of location information of the network. In practice, sensor networks often deviate from the UDG model significantly. It is not uncommon to observe stable long links that are more than five times longer than unstable short links in real wireless networks. A more general network model, the quasi unit-disk graph (quasi-UDG) model, captures much better the characteristics of wireless networks. However, the understanding of the properties of general quasi-UDGs has been very limited, which is impeding the design of key network protocols and algorithms. In this dissertation we study the properties for general wireless sensor networks and develop new topological/geometrical techniques for wireless sensor networking. We assume neither the ideal UDG model nor the location information of the nodes. Instead we work on the more general quasi-UDG model and focus on figuring out the relationship between the geometrical properties and the topological properties of wireless sensor networks. Based on such relationships we develop algorithms that can compute useful substructures (planar subnetworks, boundaries, etc.). We also present direct applications of the properties and substructures we constructed including routing, data storage, topology discovery, etc. We prove that wireless networks based on quasi-UDG model exhibit nice properties like separabilities, existences of constant stretch backbones, etc. We develop efficient algorithms that can obtain relatively dense planar subnetworks for wireless sensor networks. We also present efficient routing protocols and balanced data storage scheme that supports ranged queries. We present algorithmic results that can also be applied to other fields (e.g., information management). Based on divide and conquer and improved color coding technique, we develop algorithms for path, matching and packing problem that significantly improve previous best algorithms. We prove that it is unlikely for certain problems in operation science and information management to have any relatively effective algorithm or approximation algorithm for them.

Wireless Sensor Network (WSN) is composed of micro sensor nodes and it represents that they are small in size and cheap in cost but own limited capacity of computation and operation time. WSN is used to detect and sense events like temperature, earthquake, creature activities, atmospheric pressure and so on. By the property of wireless data transmission, WSN can be rapidly deployed and easily built up. In other hand, lifetime of WSN has been constrained by the batteries built in each sensor node. To transmit sensed data back to the base station spends the most energy for the WSN, and thus how to operate efficiently will be the key to extend the operating time of the WSN. There are a lot of related researches that proposed many routing protocols to maximize WSN lifetime and clustering is a proven routing protocol for WSN energy efficiency. The clustering method group nearby nodes together and choose one of them as a cluster-head that will transmit data back. The most important issue of clustering method is to choose which as a cluster-head. Usually, cluster-head will be chosen by probability and normal nodes will choose their own cluster-head by distance. Global-fit and Energy-Efficient (GFEE) algorithm, which is based on global-fit concept, is proposed to enhance lifetime of WSN. GFEE not only chooses cluster-head by probability and taking turns, but also bases on residual energy. All other nodes choose their cluster-head by distance and total energy consumption. Nodes with low power should be protected by some mechanisms. Experiments approved GFEE, especially in the situations of nodes widely spread or long distance transmission.

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.

This thesis considers the problem of distributed sequential detection in wireless sensor networks (WSNs), where the number of operating sensors is unknown to the fusion center. Since the energy and bandwidth of communication channel are limited in WSNs, we employ the censoring scheme in the sequential detection to achieve energy-efficiency and low communication rate. Specifically, we show by simulations that employing censoring scheme can reduce the number of local decisions that required for the fusion center to make a final decision. The results implies that the energy conservation does not necessary degrade the performance of sequential detection in terms of the expected local decisions required for making a final decisions.

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.

Thesis (M. Phil.)--University of Hong Kong, 2010. / Includes bibliographical references (leaves 93-103). Also available in print.

Thesis (Ph. D.)--Rutgers University, 2009. / "Graduate Program in Computer Science." Includes bibliographical references (p. 103-106).

Thesis (Ph. D.)--Rutgers University, 2009. / "Graduate Program in Electrical and Computer Engineering." Includes bibliographical references (p. 125-129).