In this thesis various iterative channel estimation and data detection techniques for time-varying frequency selective channels with multiple frequency offsets are proposed. Firstly, a maximum likelihood approach for the estimation of complex multipath gains (MGs) and real Doppler shifts (DSs) for a single input "single output (SISO) frequency selective channel is proposed. In a time di vision multiple access (TDMA) system, for example the third-generation global system, or mobile GSM communications, the pilot symbols are generally inadequate to provide enough resolution to estimate frequency offsets. Therefore, our approach is to use the pilot sequence for the estimation and equalization of the channel without consideration to frequency offsets, and then to use the soft estimates of the transmitted signal as a long pilot sequence to determine iteratively the multiple frequency offsets and refine the channel estimates. Inter-symbol interference (ISI) is removed with a linear structure turbo equalizer where the filter coefficients are chosen based on the minimum mean square error (MMSE) criterion. The detection performance is verified using the bit error rate (BER) curves and the frequency offset estimation performance through comparison with appropriate Cramer-Rao lower bounds. This work is then extended for a multi-user transmission system where the channel is modelled as a multi input multi output (MIMO) TDMA system. For the iterative channel estimation, the MIMO frequency selective channel is decoupled into multiple SISO flat fading sub-channels through appropriately cancelling both inter-symbol-interference (ISI) and inter-user-interference (IUI) from the received signal. The refined channel estimates and the corresponding frequency offset estimates are then obtained for each resolved MIMO multipath tap. Simulation results confirm a superior BER and estimation performance. Finally, these iterative equalization and estimation techniques are ex tended to orthogonal frequency division multiplexing (OFDM) based SISO and MIMO systems. For OFDM, the equalization is performed in two stages. In the first stage, the channel and the frequency offsets are estimated in the time domain, while in the second stage, the transmitted symbols are estimated in the frequency domain and the mean values and the variances of the symbols are determined in the frequency domain. These two procedures interact in an iterative manner, exchanging information between the time and frequency domains. Simulation studies show that the proposed iterative scheme has the ability to track frequency off sets and provide a superior BER performance as compared to a scheme that does not track frequency offsets.

In wireline multicarrier systems a cyclic prefix is generally used to facilitate simple channel equalization at the receiver. The choice of the length of the cyclic prefix is a trade-off between maximizing the length of the channel for which inter-symbol interference is eliminated and optimizing the transmission efficiency. When the length of the channel is greater than the cyclic prefix, adaptive channel shorteners can be used to force the effective channel length of the combined channel and channel shortener to be within the cyclic prefix constraint. The focus of this thesis is the design of new blind adaptive time-domain channel shortening algorithms with good convergence properties and low computational complexity. An overview of the previous work in the field of supervised partial update adaptive filtering is given. The concept of property-restoral based blind channel shortening algorithms is then introduced together with the main techniques within this class of adaptive filters. Two new partial update blind (unsupervised) adaptive channel shortening algorithms are therefore introduced with robustness to impulsive noise commonly present in wireline multicarrier systems. Two further blind channel shortening algorithms are proposed in which the set of coefficients which is updated at each iteration of the algorithm is chosen deterministically. One of which, the partial up-date single lag autocorrelation maximization (PUSLAM) algorithm is particularly attractive due to its low computational complexity. The interaction between the receiver matched filter and the channel shortener is considered in the context of a multi-input single-output environment. To mitigate the possibility of ill-convergence with the PUSLAM algorithm an entirely new random PUSLAM (RPUSLAM) algorithm is proposed in which randomness is introduced both into the lag selection of the cost function underlying SLAM and the selection of the particular set of coefficients updated at each algorithm. This algorithm benefits from robust convergence properties whilst retaining relatively low computational complexity. All algorithms developed within the thesis are supported by evaluation on a set of eight carrier serving area test loop channels.

The use of small, cheap, networked devices to collaboratively perform a task presents an attractive opportunity for many scenarios. One such scenario is the tracking and classification of an object moving through a region of interest. A single sensor is capable of very little, but a group of sensors can potentially provide a flexible, self-organising system that can carry out tasks in harsh conditions for long periods of time. This thesis presents a new framework for tracking and classification with a wire less sensor network. Existing algorithms have been integrated and extended within this framework to perform tracking and classification whilst managing energy usage in order to balance the quality of information with the cost of obtaining it. Novel improvements are presented to perform tracking and classification in more realistic scenarios where a target is moving in a non-linear fashion over a varying terrain. The framework presented in this thesis can be used not only in algorithm development, but also as a tool to aid sensor deployment planning. All of the algorithms presented in this thesis have a common basis that results from the integration of a wireless sensor network management algorithm and a tracking and classification algorithm both of which are considered state-of-the-art. Tracking is performed with a particle filter, and classification is performed with the Transferable Belief Model. Simulations are used throughout this thesis in order to compare the performance of different algorithms. A large number of simulations are used in each experiment with various parameter combinations in order to provide a detailed analysis of each algorithm and scenario. The work presented in this thesis could be of use to developers of wireless sensor network algorithms, and also to people who plan the deployment of nodes. This thesis focuses on military scenarios, but the research presented is not limited to this.

For the channel assignment problem, the adequacy of binary channel separation constraints based on the single interferer assumption and/or a constant re-use distance has been questioned by several authors. The single interferer assumption is convenient for channel assignment purposes as it leads to a generalised graph-colouring model which is simple to formulate and very popular. However, it is desirable to approximate the operational criteria more closely than a single interferer assumption model allows, by modelling the effects of multiple simultaneous interferers. This thesis addresses the problem of modelling multiple interferers in channel assignment using constraints, with a view to finding an efficient and convenient approach which offers resilience against multiple interference whilst minimising additional spectral requirements. Motivated by a discussion of the literature concerning single and multiple interference, the thesis analyses the coverage failure as progressively higher numbers of multiple simultaneous interferers occur, characterising those interferers which lead to coverage reduction. A hybrid sequential and simulated annealing heuristic is applied which obtains optimised channel assignments for analysis, created under the single interferer assumption, for two-hundred-and-forty problem cases. The library of test cases is created using a purpose-built problem generator which is applied to create problems with differing randomised distributions of transmission sites. The analysis informs the consideration of methods for the reduction/elimination of multiple interferer effects. A multiple interference model based on higher order constraints called co-channel set constraints is assessed. Results concerning the theoretical properties of these constraints, and their satisfaction, are presented. An alternative way forward is then considered, which involves challenging the commonly applied assumption that the multiple interferer assumption implies constraints are necessarily non-binary. New methods are introduced that incorporate multiple interference into the generalised graph-colouring formulation i.e. binary constraints. The methods are tested using the test problem library optimised assignments are made and their resilience against multiple interference and the spectral requirements are used to evaluate the approaches. Evidence is provided that one of the methods provides an improved model for channel assignment with multiple interference and can be recommended for use to provide constraints which perform well under the multiple objectives concerned.

An intelligent radar sensor concept has been developed using a modelling approach for prediction of sensor performance, based on application of sensor and environment models. Land clutter significantly impacts on the operation of radar sensors operating at low-grazing angles. The clutter modelling technique developed in this thesis for the prediction of land clutter forms the clutter model for the intelligent radar sensor. Fusion of remote sensing data is integral to the clutter modelling approach and is addressed by considering fusion of radar remote sensing data, and mitigation of speckle noise and data transmission impairments. The advantages of the intelligent sensor approach for predicting radar performance are demonstrated for several applications using measured data. The problem of predicting site-specific land radar performance is an important task which is complicated by the peculiarities and characteristics of the radar sensor, electromagnetic wave propagation, and the environment in which the radar is deployed. Airborne remote sensing data can provide information about the environment and terrain, which can be used to more accurately predict land radar performance. This thesis investigates how fusion of remote sensing data can be used in conjunction with a sensor modelling approach to enable site-specific prediction of land radar performance. The application of a radar sensor model and a priori information about the environment, gives rise to the notion of an intelligent radar sensor which can adapt to dynamically changing environments through intelligent processing of this a priori knowledge. This thesis advances the field of intelligent radar sensor design, through an approach based on fusion of a priori knowledge provided by remote sensing data, and application of a modelling approach to enable prediction of radar sensor performance. Original contributions are made in the areas of intelligent radar sensor development, improved estimation of land surface clutter intensity for site-specific low-grazing angle radar, and fusion and mitigation of sensor and data transmission impairments in radar remote sensing data.

A novel adaptive source-channel coding scheme for progressive transmission of medical images with a feedback system is therefore proposed in this dissertation. The overall design includes Discrete Wavelet Transform (DWT), Embedded Zerotree Wavelet (EZW) coding, Joint Source-Channel Coding (JSCC), prioritization of region of interest (RoI), variability of parity length based on feedback, and the corresponding hardware design utilising Simulink. The JSCC can achieve an efficient transmission by incorporating unequal error projection (UEP) and rate allocation. An algorithm is also developed to estimate the number of erroneous data in the receiver. The algorithm detects the address in which the number of symbols for each subblock is indicated, and reassigns an estimated correct data according to a decision making criterion, if error data is detected. The proposed system has been designed based on Simulink which can be used to generate netlist for portable devices. A new compression method called Compressive Sensing (CS) is also revisited in this work. CS exhibits many advantages in comparison with EZW based on our experimental results. DICOM JPEG2000 is an efficient coding standard for lossy or lossless multi-component image coding. However, it does not provide any mechanism for automatic RoI definition, and is more complex compared to our proposed scheme. The proposed system significantly reduces the transmission time, lowers computation cost, and maintains an error-free state in the RoI with regards to the above provided features. A MATLAB-based TCP/IP connection is established to demonstrate the efficacy of the proposed interactive and adaptive progressive transmission system. The proposed system is simulated for both binary and symmetric channel (BSC) and Rayleigh channel. The experimental results confirm the effectiveness of the design.

The advent of fourth generation (4G) wireless communication with available modulation bandwidth ranging from 1 MHz to 20 MHz is starting to emerge. The linear modulation technique being employed means that the power amplifiers that support the standards need to have high degree of linearity. By nature, however, all power amplifiers are non-linear. Load-pull measurement system provides anindispensable non-linear tool for the characterization of power amplifier and transistor for linearity enhancement. Conventional passive or active load-pull has delay problem that get worse as the modulation frequency is increased beyond few MHz. Furthermore in order to provide robust non-linear measurement, load-pull system needs to provide bandwidth at least five times the modulation bandwidth by including the fifth-order inter-modulation (IMD5). This thesis presents, for the first time, delay compensation on the unique active envelope load-pull architecture providing constant impedance for bandwidth up to 20 MHz. In doing so, it provides a superior load-pull measurement and also the ability to directly control in-band impedances. Artificial variations imposed on the in-band impedances offer further insight on power amplifier and transistor behaviours under wideband multi-tone stimulus.

Novel iterative and recursive schemes for the equalization of time-varying frequency selective channels are proposed. Such doubly selective channels are shown to be common place in mobile communication systems, for example in second generation systems based on time division multiple access (TDMA) and so-called beyond third generation systems most probably utilizing orthogonal frequency division multiplexing (OFDM). A new maximum likelihood approach for the estimation of the complex multipath gains (MGs) and the real Doppler spreads (DSs) of a parametrically modelled doubly selective single input single output (SISO) channel is derived. Considerable complexity reduction is achieved by exploiting the statistical properties of the training sequence in a TDMA system. The Cramer-Rao lower bound for the resulting estimator is derived and simulation studies are employed to confirm the statistical efficiency of the scheme. A similar estimation scheme is derived for the MGs and DSs in the context of a multiple input multiple output (MIMO) TDMA system. A computationally efficient recursive equalization scheme for both a SISO and MIMO TDMA system which exploits the estimated MGs and DSs is derived on the basis of repeated application of the matrix inversion lemma. Bit error rate (BER) simulations confirm the advantage of this scheme over equalizers which have limited knowledge of such parameters. For OFDM transmission over a general random doubly selective SISO channel, the time selectivity is mitigated with an innovative relatively low complexity iterative method. Equalization is in effect split into two stages: one which exploits the sparsity in the associated channel convolution matrix and a second which performs a posteriori detection of the frequency domain symbols. These two procedures interact in an iterative manner, exchanging information between the time and frequency domains. Simulation studies show that the performance of the scheme approaches the matched filter bound when interleaving is also introduced to aid in decorrelation. Finally, to overcome the peak to average power problem in conventional OFDM transmission, the iterative approach is extended for single carrier with cyclic prefix (SCCP) systems. The resulting scheme has particularly low complexity and is shown by simulation to have robust performance.

The focus of this thesis is to enhance the performance of wireless communication systems through the exploitation of multiple antennas at both the transmitter and the receiver ends of a communication link. Such a multiple-input multiple-output (MIMO) connection can theoretically provide spatially independent channels which can be exploited to provide diversity gain and thereby mitigate the problem of channel fading. To integrate such MIMO technology with emerging wireless systems such as third generation code division multiple access (CDMA) and fourth generation orthogonal division multiple access (OFDMA) based-approaches novel advanced signal processing techniques are required. The major advantages of MIMO systems, including array, diversity and multiplexing gains, are initially reviewed. Diversity gain is identified as the key property, which leverages the spatial independent channels to increase the robustness of the communication link. The family of space-time block codes is then introduced as a low computational complexity scheme to benefit from diversity gain within wireless systems. In particular, extended-orthogonal and quasi-orthogonal space-time block codes (EO-/QO-STBCs) are introduced for systems with four transmit antennas which can operate either in open or closed-loop forms. New EO-STBC and QO-STBC wideband CDMA transmission schemes are proposed which when operating in closed-loop mode, i.e. channel state information is exploited at the transmitter, is shown to attain full diversity and thereby outperform previous schemes in terms of attain able symbol error rate performance. This advantage is then utilized in MIMO-OFDM transmission schemes and similar frame error rate (FER) performance advantage is attained. Finally, to mitigate multiuser interference within the proposed MIMO-OFDM system a novel two-step combined parallel interference canceller and multiuser detection scheme is proposed. Simulation studies based upon FER confirm the efficacy of the technique.

Network operators desire effective, pragmatic solutions to instances of the cell planning problem in order to improve their quality of service, enhance network coverage and capacity capability, and ultimately increase company profits. Previ ous cell plans have been constructed manually but these methods do not produce the best network configuration. More reliance has since been placed on automated cell planning to produce effective solutions. The introduction of the Universal Mobile Telecommunication System (UMTS) emphasizes the need for high perfor mance planning tools. Motivated by a discussion of the literature concerning cell planning, an existing model for Global System for Mobile Communication (GSM) is modified to take account of the requirements of UMTS networks. A suite of test cases is created using a purpose-built problem generator, including problems with a range of site and traffic distributions for rural, suburban and urban markets. Traditionally, cell planning has been seen purely as an optimisation problem, neglecting the pre-optimisation stage of network dimensioning. This thesis inves tigates the effect of network dimensioning as a precursor to optimisation demon strating the benefits of cell planning in three stages consisting of site estimation, site selection and optimisation. The first stage, site estimation, utilises previously published lower bounding techniques to provide a means of approximating the number of sites required to meet capacity targets in the uplink and downlink. Site selection compares random selection to three newly developed algorithms to make effective automatic selections of sites from a candidate set. The final optimiza tion phase presents a framework based on the tabu search meta-heuristic capable of optimising the dimensioned network designs with respect to the representative operational scenarios. Multiple traffic snapshot evaluations are considered in the optimisation objective function.