Wireless sensor networks (WSNs) have been developed for a variety of appli- cations such as battle�eld surveillance, environment monitoring, health care and so on. For such applications, the design of WSN has been limited by two main resource constraints, power and delay. Therefore, since wireless sensors with a small battery are subject to strict power constraints, the e�cient usage of power is one of the im- portant challenges. As delay-sensitive applications are emerging, they have been in demand for making a quick decision with the enhanced detection accuracy. Under above constraints, we propose a sequential detection scheme and compare it with a Fixed-sample-size (FSS) detection scheme in terms of power and delay. Our main contribution is to analyze the overall system performance of the proposed scheme in the statistical signal processing framework under of power and delay constraints. In this thesis, we evaluate the overall system performance of sequential detection scheme in a serial multi-hop WSN topology. For sequential detection, the sensor nodes continue to relay the observations to the next node until the sequential detector makes a �nal decision based on the observations. On the other hand, the FSS detector waits until all the observations come to the fusion center, and then gives a �nal decision. For a fair comparison of the two schemes with respect to power and delay, the initial step is to �nd the same detection performance region between the two schemes. Detection performance is evaluated with performance measures such as false alarm, miss and prior probability. Simulation results show that each scheme has an advantage and a disadvantage concerning power and delay respectively. That is, sequential detection performs more e�ciently in delay since the number of samples in sequential detection is less on average than in FSS detection to obtain the same detection performance. However, FSS detection with a small number of packet paths consumes less power than sequential detection. Through the analysis of a cost function, which is a linear combination of power and delay, we compare the cost value between the two schemes and �nd less region of the cost value in both schemes. This analysis will provide a good starting point and foundation for designing an e�cient multi-hop WSN with small power and delay constraints.

In the literature, most distributed detection developed so far mainly focuses on the test rule based on fixed sample size. However, in the real situations, sequential tests are more suitable to be utilized since it might achieve the same detection performance by using fewer number of samples as compared with the fixed-sample-size test. Thus, this theses will propose a new distributed sequential detection approach for the applications in wireless sensor networks(WSNs) and cognitive radios(CRs). First we refer to the sequential detection, and it has been developed by Wald in 1994, which is well known as the sequential probability ratio test (SPRT). The SPRT is proved to be able to decrease the required average sample numbers or reducing the average detection time. Indeed, the SPRT is the optimal sequential detection in terms of the minimizing the required number of samples given the constraint of false alarm and miss probabilities when the observation samples are independent and identical distributed (i.i.d.). However, if the observation samples are not dentically distributed, by simulation results show that the SPRT is not the optimal test. Based on a heuristic approach, this thesis then developed a new distributed detection scheme based on the sorted samples. Finally , the simulation results obtained by this thesis shows that the proposed scheme can further reduce the number of samples required for making the final decision as compared with SPRT.

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.

To combat channel fading, cooperative communication achieves spatial diversity for the transmission between source and destination through the help of relay. However, if the relay behaves abnormally or maliciously and the destination is not aware, the diversity gain of the cooperative system will be significantly reduced, which degrades system performance. In our thesis, we consider an one-relay decode and forward cooperative network, and we assume that the relay may misbehave with a certain probability. If the relay is malicious, it will garble transmission signal, resulting in severe damage to cooperative system. In this work, we discuss three kinds of malicious behavior detection. More specifically, we adopt sequential detection to detect the behavior of relay. If tracing symbols are inserted among the source message, the destination detects malicious after extracting the received tracing symbols. We adopt log-likelihood ratio test to examine these tracing symbols, and then determine the behavior of relay. If the source does not transmit tracing symbols, the destination detects misbehavior according to the received data signal. Furthermore, we employ sequential detection to reduce detection time for a given probabilities of false alarm and miss detection. Through simulation results, for a certain target on probability of errors, our proposed methods can effectively reduce numbers of observations. On the other works, the destination can effectively detect misbehavior of relay, and eliminating the damage causes by malicious relay without requiring large numbers of observations.

Traditionally radio spectrum was considered a commodity to be allocated in a fixed and centralized manner, but now the technical community and the regulators approach it as a shared resource that can be flexibly and intelligently shared between competing entities. In this thesis we focus on novel strategies to sense and access the radio spectrum within the framework of Opportunistic Spectrum Access via Cognitive Radio Networks (CRNs). In the first part we develop novel transmit opportunity detection methods that effectively exploit the gray space present in packet based networks. Our methods proactively detect the maximum safe transmit power that does not significantly affect the primary network nodes via an implicit feedback mechanism from the Primary network to the Secondary network. A novel use of packet interarrival duration is developed to robustly perform change detection in the primary network's Quality of Service. The methods are validated on real world IEEE 802.11 WLANs. In the second part we study the inferential use of Goodness-of-Fit tests for spectrum sensing applications. We provide the first comprehensive framework for decision fusion of an ensemble of goodness-of-fit tests through use of p-values. Also, we introduce a generalized Phi-divergence statistic to formulate goodness-of-fit tests that are tunable via a single parameter. We show that under uncertainty in the noise statistics or non-Gaussianity in the noise, the performance of such non-parametric tests is significantly superior to that of conventional spectrum sensing methods. Additionally, we describe a collaborative spatially separated version of the test for robust combining of tests in a distributed spectrum sensing setting. In the third part we develop the sequential energy detection problem for spectrum sensing and formulate a novel Sequential Energy Detector. Through extensive simulations we demonstrate that our doubly hierarchical sequential testing architecture delivers a significant throughput improvement of 2 to 6 times over the fixed sample size test while maintaining equivalent operating characteristics as measured by the Probabilities of Detection and False Alarm. We also demonstrate the throughput gains for a case study of sensing ATSC television signals in IEEE 802.22 systems. / text

Land cover classification using remotely sensed data is a critical first step in large-scale environmental monitoring, resource management and regional planning. The classification task is made difficult by severe atmospheric scattering and absorption, seasonal variation, spatial dependence, complex surface dynamics and geometries, and large intra-class variability. Most of the recent research effort in land cover classification has gone into the development of increasingly robust and accurate (and also increasingly complex) classifiers by constructing–often in an ad hoc manner–multispectral, multitemporal, multisource classifiers using modern machine learning techniques such as artificial neural networks, fuzzy-sets, and expert systems. However, the focus has always been (almost exclusively) on increasing the classification accuracy of newly developed classifiers. We would of course like to perform land cover classification (i) as accurately as possible, but also (ii) as quickly as possible. Unfortunately there exists a tradeoff between these two requirements, since the faster we must make a decision, the lower we expect our classification accuracy to be, and conversely, a higher classification accuracy typically requires that we observe more samples (i.e., we must wait longer for a decision). Sequential analysis provides an attractive (indeed an optimal) solution to handling this tradeoff between the classification accuracy and the detection delay–and it is the aim of this study to apply sequential analysis to the land cover classification task. Furthermore, this study deals exclusively with the binary classification of coarse resolution MODIS time series data in the Gauteng region in South Africa, and more specifically, the task of discriminating between residential areas and vegetation is considered. / Dissertation (MEng)--University of Pretoria, 2011. / Electrical, Electronic and Computer Engineering / unrestricted

This dissertation focuses on real-time decision-making for large-scale domain search and object classification using Multiple Autonomous Vehicles (MAV). In recent years, MAV systems have attracted considerable attention and have been widely utilized. Of particular interest is their application to search and classification under limited sensory capabilities. Since search requires sensor mobility and classification requires a sensor to stay within the vicinity of an object, search and classification are two competing tasks. Therefore, there is a need to develop real-time sensor allocation decision-making strategies to guarantee task accomplishment. These decisions are especially crucial when the domain is much larger than the field-of-view of a sensor, or when the number of objects to be found and classified is much larger than that of available sensors. In this work, the search problem is formulated as a coverage control problem, which aims at collecting enough data at every point within the domain to construct an awareness map. The object classification problem seeks to satisfactorily categorize the property of each found object of interest. The decision-making strategies include both sensor allocation decisions and vehicle motion control. The awareness-, Bayesian-, and risk-based decision-making strategies are developed in sequence. The awareness-based approach is developed under a deterministic framework, while the latter two are developed under a probabilistic framework where uncertainty in sensor measurement is taken into account. The risk-based decision-making strategy also analyzes the effect of measurement cost. It is further extended to an integrated detection and estimation problem with applications in optimal sensor management. Simulation-based studies are performed to confirm the effectiveness of the proposed algorithms.

This study aims to present a sequential method for the classification of the autoregressive processes. Different from the conventional detectors having fixed sample size, the method uses Wald&rsquo / s sequential probability ratio test and has a variable sample size. It is shown that the suggested method produces the classification decisions much earlier than fixed sample size alternative on the average. The proposed method is extended to the case when processes have unknown variance. The effects of the unknown process variance on the algorithmperformance are examined. Finally, the suggested algorithm is applied to the classification of fixed and rotary wing targets. The average detection time and its relation with signal to noise ratio are examined.

Cognitive Radios are emerging communication systems which efficiently utilize the unused licensed radio spectrum called spectral holes. They run Spectrum sensing algorithms to identify these spectral holes. These holes need to be identified at very low SNR (<=-20 dB) under multipath fading, unknown channel gains and noise power. Cooperative spectrum sensing which exploits spatial diversity has been found to be particularly effective in this rather daunting endeavor. However despite many recent studies, several open issues need to be addressed for such algorithms. In this thesis we provide some novel cooperative distributed algorithms and study their performance. We develop an energy efficient detector with low detection delay using decentralized sequential hypothesis testing. Our algorithm at the Cognitive Radios employ an asynchronous transmission scheme which takes into account the noise at the fusion center. We have developed a distributed algorithm, DualSPRT, in which Cognitive Radios (secondary users) sequentially collect the observations, make local decisions and send them to the fusion center. The fusion center sequentially processes these received local decisions corrupted by Gaussian noise to arrive at a final decision. Asymptotically, this algorithm is shown to achieve the performance of the optimal centralized test, which does not consider fusion center noise. We also theoretically analyze its probability of error and average detection delay. Even though DualSPRT performs asymptotically well, a modification at the fusion node provides more control over the design of the algorithm parameters which then performs better at the usual operating probabilities of error in Cognitive Radio systems. We also analyze the modified algorithm theoretically. DualSPRT requires full knowledge of channel gains. Thus we extend the algorithm to take care the imperfections in channel gain estimates. We also consider the case when the knowledge about the noise power and channel gain statistic is not available at the Cognitive Radios. This problem is framed as a universal sequential hypothesis testing problem. We use easily implementable universal lossless source codes to propose simple algorithms for such a setup. Asymptotic performance of the algorithm is presented. A cooperative algorithm is also designed for such a scenario. Finally, decentralized multihypothesis sequential tests, which are relevant when the interest is to detect not only the presence of primary users but also their identity among multiple primary users, are also considered. Using the insight gained from binary hypothesis case, two new algorithms are proposed.

Opportunistic use of unused spectrum could efficiently be carried out using the paradigm of Cognitive Radio (CR). A spectrum remains idle when the primary user (licensee) is not using it. The secondary nodes detect this spectral hole quickly and make use of it for data transmission during this interval and stop transmitting once the primary starts transmitting. Detection of spectral holes by the secondary is called spectrum sensing in the CR scenario. Spectrum Sensing is formulated as a hypothesis testing problem wherein under H0 the spectrum is free and under H1, occupied. The samples will have different probability distributions, P0 and P1, under H0 and H1 respectively. In the first part of the thesis, a new algorithm - entropy test is presented, which performs better than the available algorithms when P0 is known but not P1. This is extended to a distributed setting as well, in which different secondary nodes collect samples independently and send their decisions to a Fusion Centre (FC) over a noisy MAC which then makes the final decision. The asymptotic optimality of the algorithm is also shown. In the second part, the spectrum sensing problem under impediments such as fading, electromagnetic interference and outliers is tackled. Here the detector does not possess full knowledge of either P0 or P1. This is a more general and practically relevant setting. It is found that a recently developed algorithm (which we call random walk test) under suitable modifications works well. The performance of the algorithm theoretically and via simulations is shown. The same algorithm is extended to the distributed setting as above.