<p>The rapid development of fiber-optic communication system requires a increasing transmission data rate and reach. One of the challenge for long-haul high-speed fiber-optic system is how to reduce the impairments and degrading effects from fiber dispersions and nonlinearity. Although the advances of digital signal processor (DSP) make impairment compensation for the coherent systems become possible, the implementation of coherent system is still expensive. Among the direct detection systems, differential phase-shift keying (DPSK) system shows great advantages over other modulation schemes. Although the recent commercial fiber-optic communication systems are based on DPSK, polarization-mode dispersion and fiber nonlinearity are still limiting factors for DPSK systems. In 2004, a differential polarization-phase-shift keying (DPolPSK) system in which information is encoded in both polarization and phase with multi-level direct detection is proposed, and it is found out the DPolPSK system greatly reduces the effect of nonlinear polarization scattering. With the expectation of getting a better nonlinear tolerance than DPSK system, a novel differential polarization-shift keying (DPolSK) system with balanced direct detection receiver is proposed and compared with DPSK system in terms of fiber dispersion and nonlinearity tolerance.</p> <p>The DPolSK system encodes information in the polarization angle difference between two adjacent symbols. To transmit bit '1' ('0'), the polarization angle of the current symbol is shifted by π/2 [-/2 (-π/2) with respect to the previous symbol. A balanced detector based on optical delay interferometer and Faraday rotator is used to demodulate the DPolSK signals. Ideally without dispersion and nonlinearity, the DPolSK has the same receiver current level as DPSK system.</p> <p>Monter-carlo simulations are conducted to evaluate the DPolSK system in the presence of chromatic dispersion (CD), polarization-mode dispersion (PMD), and fiber nonlinearity (FNL). The simulation results show DPolSK system has the same bit-error-rate (BER) and PMD tolerance as DPSK system when fiber nonlinearity is ignored. However, when nonlinearity is taken account in the system, DPolSK shows overall superiority to DPSK system. The reason that DPolSK has higher nonlinear tolerance than DPSK is also explored. The intra-channel four-wave mixing (IFWM) effect on the DPolSK and DPSK systems are investigated and the simulation results show the ghost pulse generation induced by IFWM is suppressed in DPolSK system as compared to DPSK system, leading to higher nonlinear tolerance.</p> / Master of Applied Science (MASc)

<p>Model-based approaches to 3D object tracking and pose estimation that employ a particle filter are effective and robust , but computational complexity limits their efficacy in real-time scenarios. This thesis describes a novel framework for acceleration of particle filtering approaches to 3D model-based, markerless visual tracking in monocular video using a graphics processing unit (GPU). Specifically, NVIDIA compute unified device architecture (CUDA) and Direct3D are employed to harness the single-instruction multiple- thread (SHvIT) programming model used by the GPU's massively parallel streaming multiprocessors (SMs) for simulation (3D model rendering) and evaluation (segmentation, feature extraction, and weight calculation) of hundreds of particles at high speeds. The proposed framework addresses the computational intensity that is intrinsic to all particle filter approaches, including those with modifications and extensions that strive to reduce the number of required particles while maintaining tracking quality.</p> <p>The sampling importance resampling (SIR) particle filter and its utility in 3D model-based tracking is reviewed and a detailed overview of relevant GPU-programming concepts is presented. The proposed framework is formulated as a series of interconnected steps and the functionality and implementation of each is described in det ail. Rigid and articulated tracking examples are presented in the context of human-computer interact ion (HCI) and augmented reality (AR) applications, with a focus on bare hand tracking. Performance and tracking quality results demonstrate markerless, model-based visual t racking on consumer-grade hardware with pixel-level accuracy up to 95 percent at 30+ frames per second. The framework accelerates particle evaluation up to 25 times over a comparable CP -only implementation. providing an increased particle count while maintaining real-time frame rates.</p> / Master of Applied Science (MASc)

<p>Track-Before-Detect (TBD) algorithms are far more efficient over standard DetectBefore- Track (DBT) target tracking approach for tracking targets in low Signal-toNoise- Ratio (SNR) environment . With low SNR scenario the target amplitude may never be strong enough to exceed threshold value and under classical setting such cases will not lead to detection. This might be the case in spatially diversified multiple sensors network like Multiple-Input-Multiple-Output (MIMO) radars. Through letting the tracking directly on the unthresholded data, TBD techniques exploit all the information in the received measurement signal to yield detection and tracking simultaneously. With TBD framework an efficient multitarget, non-linear filtering algorithm is an issue to extract information from target dynamics. In t his thesis Probability-Hypothesis-Density (PHD) filter implementation of a recursive TBD algorithm is proposed. The PHD filter, propagating only the first-order statistical moment of the full target posterior, is a computationally efficient solution to multitarget tracking problems with varying number of targets. Furthermore a PHD filter based tracking algorithm avoids the preassumption of the maximum number of targets performing the state estimation together with number of targets.</p> <p>The PHD filter based TBD algorithm is applied to multitarget tracking with MIMO Radars. With widely-separated transmitters and receivers of MIMO system the Radar-Cross-Section (RCS) diversity can be utilized by illuminating the target from ideally uncorrelated aspects. Multiple sensor TBD is proposed in order to process the measurement signals from different multiple transmitter-receiver pairs in the MIMO Radar system. In this model target observability to the sensor as a result of target RCS diversity is taken in to consideration in the likelihood calculation. In order to provide a benchmark for testing the proposed algorithm performance, Posterior Cramer-Rao Lower Bound (PCRLB) for widely-separated MINIO radar is also presented. Monte Carlo simulations have been done on multitarget scenarios with various SNR values and target motion models. Performance evaluation on simulation results demonstrates the improved performance of the proposed tracking algorithm.</p> / Master of Applied Science (MASc)

<p>Collaborative haptic simulations allow multiple users in a virtual environment to simultaneously interact with shared virtual objects. The implementation of shared virtual environments over a network removes the geographical barriers and enables users from around the globe to modify the environment and in addition feel the presence of other users. However, network issues arise in the communication of data over a network such as the Internet. Communication channel delay, jitter, packet loss and limited packet transmission rate can adversely affect the performance of collaborative haptic systems and may even cause instability.</p> <p>This thesis builds upon our group's recent work in distributed networked haptics (Fotoohi <em>et al</em>. [31]). The proposed distributed peer-peer architecture improved the haptic simulation performance over the centralized architecture by providing local high-rate feedback to the users in a Local Area Network (LAN). Virtual spring-damper couplers synchronized the multiple copies of the virtual environment and coupled the users to the virtual objects.</p> <p>Forming on the above distributed architecture, this thesis proposes methods for improving the performance and stability of shared haptic environments with a stronger emphasis on the effects of time delay in the context of Internet communication. To this end new quantitative measures are presented for quantifying the fidelity of haptic simulations in such environments. User's perceived admittance and discrepancy among local copies of virtual objects are considered in defining these measures. Furthermore, state prediction and feedforward schemes are proposed to compensate for the negative effects of the network communication delay on the transparency and stability of the haptic simulation. An optimization problem is formulated for selecting the virtual coupling gains that can enhance the performance while maintaining system stability. The solution to the this problem provides us with the set of control parameters that optimize the defined performance measures.</p> <p>A three user distributed architecture is presented to show the extension of the proposed methods to haptic simulations involving more than two users. Numerical analysis and haptic interaction experiments over the Internet are carried out to demonstrate the effectiveness of the proposed approach in two-user and three-user platforms. The obtained analytical and experimental results verified improvements by the prediction and feedforward mechanisms.</p> / Master of Applied Science (MASc)

<p>In this thesis, Robust Slepian-Wolf coding problem is discussed. Two correlated source sequences Xⁿ and Yⁿ are encoded at separate encoders and decoded together. When the encoder of source Y is broken, another sequence Xⁿ still can be decoded to achieve a nontrivial distortion. Further, Xⁿ can be recovered losslessly once that broken encoder is restored. A practical coding scheme is developed using low density graph codes. Moreover, by generalizing the coding scheme of Robust Slepian-Wolf coding problem, two approaches are proposed for the Wyner-Ziv problem using the low density graph codes.</p> / Master of Applied Science (MASc)

<p>In this thesis, we reexamine the classical problems of image/video spatial resolution up conversion and video deinterlacing with an aim to develop real-time, adaptive solutions. The research of this thesis is important because most video applications require real time throughput. We study the use of GPU (Graphics Processing Unit) technology for high throughput video interpolation and deinterlacing. The main technical challenge is how to fully utilize the processing power and parallel architecture of GPU to maximize the throughput of up conversion and deinterlacing without compromising the visual quality of the resulting videos. To achieve the goal we develop a GPU-friendly two-pass directional image/video resolution up conversion algorithm and present a GPU implementation of the method, using the NVIDIA CUDA (Compute Unified Device Architecture) technology. We also devise a GPU-motivated motion-adaptive deinterlacing algorithm and develop a CUDA-based implementation of the algorithm. To strike a balance between performance and complexity, we discuss the techniques of adapting the computations in motion detection and adaptive directional interpolation to the GPU architecture for maximum video throughput possible. Experimental results demonstrate that using a mid-range GPU card, our CUDA-based implementations offer real-time solutions for image/video spatial resolution upconversion and video deinterlacing.</p> / Master of Applied Science (MASc)

<p>A Sinusoidal Modulation technique is proposed for the synthetic loading of induction machines at full load based on power electronics control. The sinusoidal modulations are based on the two axis method and reference-frame theory. The method applies specifically to induction machines for which physical load testing is unadvisable or unattainable. This allows motor manufactures to fully test their products while utilizing current motor production as the basic setup of their test system decreasing costs.</p> <p>Previous modulation technique methods are either too costly or incur high oscillation currents rendering the systems impractical. The Sinusoidal Modulation technique computer simulations show the ability of the method to suppress power fluctuations on the driver motor of the system. Experimental and simulation results are analyzed and compared to prove simulation accuracy and system viability. Experimental data gathered at half rated load show maximum fluctuations on the driver motor in the range of 2%.</p> / Master of Applied Science (MASc)

<p>Traditional wireless sensor networks (WSNs) working in the license-free spectrum suffer from uncontrolled interference as the license-free spectrum becomes increasingly crowded. Designing a WSN based on cognitive radio can be promising in the near future in order to provide data transmissions with quality of service requirements. In this thesis, we introduce a cognitive radio sensor network (CRSN) and analyze its performance for supporting real-time traffic. The network devices opportunistically access vacant (or available) channels in the licensed spectrum. When the current channel becomes unavailable, the devices can switch to a new channel.</p> <p>Three types of real-time traffic are considered, constant-bit-rate (CBR) traffic, bursty traffic, and Poisson traffic. For the CBR traffic, a fixed number of packets are generated periodically; for the bursty traffic, a burst of packets are generated periodically and the number of packets in each burst is random; and for the Poisson traffic, the packet arrivals follow Poisson process. We derive the average packet transmission delay for each type of the traffic. The analytical results are verified by computer simulations. Our results indicate that real-time traffic can be effectively supported in the CRSN, and packets with the Poisson arrivals may experience longer average delay than the bursty arrivals.</p> / Master of Applied Science (MASc)

<p>In this thesis, we propose a new encoder-friendly image compression strategy for high-throughput cameras and other scenarios of resource-constrained encoders. The encoder performs L_∞-constrained predictive coding (DPCM coupled with uniform scalar quantizer), while the decoder solves an inverse problem of L₂ restoration of L_∞-coded images. Although designed for minimum encoder complexity (lower than distributed source coding and compressive sensing), the new codec outperforms state-of-the-art encoder-centric image codecs such as JPEG 2000 in PSNR for bit rates higher than 1.2 bpp, while maintaining much tighter L_∞ error bounds as well. This is achieved by exploiting the tight error bound on each pixel provided by the L_∞-constrained encoder and by locally adaptive image modeling.</p> / Master of Applied Science (MASc)

<p>Multi-input multi-output (MIMO) technology has been used to improve wireless communications systems over the past several years. The multiple antennas of MIMO systems are used to increase data rates through multiplexing gain and/or increase the reliability of the system through diversity gain. It is known that an optimum trade-off between diversity gain and multiplexing gain can be achieved by having proper space-time block code (STBC) designs. The current STBC designs minimizing the pair-wise error probability (PEP) of the maximum likelihood (ML) detector are based mainly on the rank and the determinant criteria.</p> <p>In this thesis, we study a special case of the MIMO system. This consists of a coherent communication system equipped with Mt transmitter antennas and a single receiver antenna, i.e., a multi-input single-output (MISO) system. Such systems are often encountered in mobile down-link communications for which a MIMO realization may be expensive or for which the mobile receiver may not be able to support multiple antennas (e.g. a mobile phone). Given a full-rate data transmission, the PEP of the ML detector for such systems can be minimized by using the rotated quasi-orthogonal STBC design, which enables full diversity and optimal coding gains for the system. The efficiency of fast ML decoding for orthogonal STBC is also largely preserved for such quasi-orthogonal STBC. However, for large constellations, the performance of such "optimum" codes deteriorates due to the increase of the number of the nearest neighbours per symbol.</p> <p>To correct such a deficiency of the code, in this thesis, we propose to include the number of nearest neighbors in the design criterion. We show that for the current optimal rotated quasi-orthogonal code, the number of nearest neighbours tends to infinity when the size of constellation becomes infinite. However, we show that by having a particular value of rotation, not only full diversity and maximum coding gain will be achieved, but also a small number of nearest neighbours will be maintained even for very large constellations.</p> <p>Also, at present, STBC designs in a MIMO system are mainly based on the PEP (or its upper bound). This is because the geometrical structure of the decision regions for a general MIMO channel, equipped with the ML detector, is so irregular that it would be impossible to obtain an exact error probability formula for the ML receiver. This means that the error probability formula cannot be utilized as a criterion for the design of the optimal transmitter for the MIMO systems and the current STBC designs may not be truly optimum in terms of the exact error probability. To rectify this problem, in this thesis, we first find a closed form algorithm for ML detection, for a 4 x 1 MISO system equipped with a ML detector, transmitting signals from a four signal quadrature amplitude modulation (4-QAM) constellation. This algorithm is derived such that given a received signal and the channel, the transmitted signal can be obtained using a threshold decision. Then, using this ML detection algorithm, a closed form of the exact error probability of this system is derived. This closed form decoding algorithm is then applied to the 4-group decodable STBC, in order to obtain the optimal rotation angle to minimize the ML error probability.</p> / Master of Applied Science (MASc)