The thesis proposes a third order continuous-time sigma delta modulator used in GSM. We used a special 1.5bit quantizer, and to use its three different states to reach a differential feedback path. That can improve the resolution of our circuit. Oversampling and noise shaping are two keys of sigma delta modulator. In structure, the continuous-time features can reduce power consumption. The proposed sigma delta modulator uses TSMC 0.35 m CMOS process and its sampling frequency is 10.8MHz, bandwidth is200KHz and oversampling ratio is 32.

Nowadays, the multi-standard wireless receivers and multi-format video processors have created a great demand for integrating multiple standards into a single chip. The multiple standards usually require several Analog to Digital Converters (ADCs) with different specifications. A promising solution is adopting a power and area efficient reconfigurable ADC with tunable bandwidth and dynamic range. The advantage of the reconfigurable ADC over customized ADCs is that its power consumption can be scaled at different specifications, enabling optimized power consumption over a wide range of sampling rates and resulting in a more power efficient design. Moreover, the reconfigurable ADC provides IP reuse, which reduces design efforts, development costs and time to market. On the other hand, software radio transceiver has been introduced to minimize RF blocks and support multiple standards in the same chip. The basic idea is to perform the analog to digital (A/D) and digital to analog (D/A) conversion as close to the antenna as possible. Then the backend digital signal processor (DSP) can be programmed to deal with the digital data. The continuous time (CT) bandpass (BP) sigma-delta ADC with good SNR and low power consumption is a good choice for the software radio transceiver. In this work, a proposed 10-bit reconfigurable ADC is presented and the non-overlapping clock generator and state machine are implemented in UMC 90nm CMOS technology. The state machine generates control signals for each MDAC stage so that the speed can be reconfigured, while the power consumption can be scaled. The measurement results show that the reconfigurable ADC achieved 0.6-200 MSPS speed with 1.9-27 mW power consumption. The ENOB is about 8 bit over the whole speed range. In the second part, a 2-bit quantizer with tunable delay circuit and 2-bit DACs are implemented in TSMC 0.13um CMOS technology for the 4th order CT BP sigma-delta ADC. The 2-bit quantizer and 2-bit DACs have 6dB SNR improvement and better stability over the single bit quantizer and DACs. The penalty is that the linearity of the feedback DACs should be considered carefully so that the nonlinearity doesn't deteriorate the ADC performance. The tunable delay circuit in the quantizer is designed to adjust the excess loop delay up to +/- 10% to achieve stability and optimal performance.

The design of analog and complex mixed-signal circuits in a deep submicron CMOS process technology is a big challenge. This makes it desirable to shift data converter design towards the digital domain. The advantage of using a fully digital ADC design rather than a traditional analog ADC design is that the circuit is defined by an HDL description and automatically synthesized by tools. It offers low power consumption, low silicon area and a fully optimized gate-level circuit that reduces the design costs, etc. The functioning of an all-digital ADC is based on the time domain signal processing approach, which brings a high time resolution obtained by the use of a nanometer CMOS process. An all-digital ADC design is implemented by using a combination of the digital Voltage-Controlled Oscillator (VCO) and a Time-to-Digital Converter (TDC). The VCO converts the amplitude-domain analog signal to a phase-domain time-based signal. In addition, the VCO works as a time based quantizer. The time-based signal from the VCO output is then processed by the TDC quantizer in order to generate the digital code sequences. The fully digital VCO-based ADC has the advantage of superior time resolution. Moreover, the VCO-based ADC offers a first order noise shaping property of its quantization noise. This thesis presents the implementation of a VCO-based ADC in STM 65 nm CMOS process technology using digital tools such as ModelSim simulator, Synopsys Design Compiler and Cadence SOC Encounter. The circuit level simulations have been done in Cadence Virtuoso ADE. A multi-phase VCO and multi-bit quantization architecture has been chosen for this 8-bit ADC. The power consumption of the ADC is approximately 630 μW at 1.0 V power supply and the figure of merit is around 410 fJ per conversion step.

This thesis addresses the design of distributed scalar quantizers (DSQs) for two sensors, tailored to maximize the classification accuracy for a pre-trained binary linear classifier at the central node, diverging from traditional designs that prioritize data reconstruction quality. The first contribution of this thesis is the development of efficient globally optimal DSQ design algorithms for two correlated discrete sources when the quantizer cells are assumed to be convex. First, it is shown that the problem is equivalent to a minimum weight path problem (with certain constraints) in a weighted directed acyclic graph. The latter problem can be solved using dynamic programming with O(K_1K_2M^4) computational complexity, where Ki, is the number of cells for the quantizer of source i, i = 1, 2, and M is the size of the union of the sources’ alphabets. Additionally, it is proved that the dynamic programming algorithm can be expedited by a factor of M by exploiting the so called Monge property, for scenarios where the pre-trained classifier is the optimal classifier for the unquantized sources. Next, the design of so-called staggered DSQs (SDSQs) is addressed, i.e., DSQ’s with K_1 = K_2 = K and with the thresholds of the two quantizers being interleaved. First, a faster dynamic programming algorithm with only O(KM^2) time complexity is devised for the design of the SDSQ that minimizes an upperbound on the classification error. This sped up is obtained by simplifying the graph model for the problem. Moreover, it is shown that this algorithm can also be further accelerated by a factor of M when the pre-trained linear classifier is the optimal classifier. Furthermore, some theoretical results are derived that provide support to imposing the above constraints to the DSQ design problem in the case when the pre-trained classifier is optimal. First, it is shown that when the sources (discrete or continuous) satisfy a certain symmetry property, the SDSQ that minimizes the modified cost also minimizes the original cost within the class of DSQs without the staggerness constraint. For continuous sources, it is also shown that the SDSQ that minimizes the modified cost also minimizes the original cost and all quantizer thresholds are distinct, even if the sources do not satisfy the aforementioned symmetry condition. The latter result implies that DSQs with identical encoders are not optimal even when the sources has the same marginal distribution, a fact which is proved here for the first time, up to our knowledge. The last (but not least) contribution of this thesis resides in leveraging the aforementioned results to obtain efficient globally optimal solution algorithms for the problem of decentralized detection under the probability of error criterion of two discrete vector sources that are conditionally independent given any class label. The previously known globally optimal solution has O(N^(K_1+K_2+1)) time complexity, where N is the size of the union of the alphabets of the two sources. We show that by applying an appropriate transformation to each vector source, the problem reduces to the problem of designing the optimal DSQ with convex cells in the transformed scalar domain for a scenario where the pre-trained linear classifier is the optimal classifier. We conclude that the problem can be solved by a much faster algorithm with only O(K_1K_2N^3) time complexity. Similarly, for the case of equal quantizer rates, the problem can be solved in O(KN) operations if the sources satisfy an additional symmetry condition. Furthermore, our results prove the conjecture that for continuous sources, imposing the constraint that the encoders be identical precludes optimality, even when the marginal distributions of the sources are the same. / Thesis / Doctor of Philosophy (PhD)

With low bit rate scenarios, a hybrid video coder (e.g. AVC/H.264) tends to allocate greater portion of bits for motion vectors, while saving bits on residual errors. According to this fact, a coding scheme with non-normative global motion models in combination with conventional local motion vectors is proposed, which describes the motion of a frame by the affine motion parameter sets drawn by motion segmentation of the luminance channel. The motion segmentation task is capable of adapting the number of motion objects to the video contents. 6-D affine model sets are driven by linear regression from the scalable block-based motion fields estimated by the existent MPEG encoder. In cases that the number of motion objects exceeds a certain threshold, the global affine models are disabled. Otherwise the 4 scaling factors of the affine models are compressed by a vector quantizer, designed with a unique cache memory for efficient searching and coding. The affine motion information is written in the slice header as a syntax. The global motion information is used for compensating those macroblocks whose Lagrange cost is minimized by the AFFINE mode. The rate-distortion cost is computed by a modified Lagrange equation, which takes into consideration the perceptual discrimination of human vision in different areas. Besides increasing the coding efficiency, the global affine model manifests the following two features that refine the compressed video quality. i) When the number of slices per frame is more than 1, the global affine motion model can enhance the error-resilience of the video stream, because the affine motion parameters are duplicated in the headers of different slices over the same frame. ii) The global motion model predicts a frame by warping the whole reference frame and this helps to decrease blocking artifacts in the compensation frame.

This paper presents a distributed feedback and feedforward of discrete-time delta sigma modulator applications in the radio. We know the delta-sigma modulator using oversampling and noise shaping technique, thus we can relax the specifications of the components. This paper described the architectural differences and compare, the in-band signal is less sensitive to noise interference, and improve the resolution of the circuit. In the resonator, a simple structure with a small number of capacitor in resonator circuit. This paper uses the TSMC 0.18£gm process parameters to the simulation, implementation, and measurement. Our fourth-order discrete-time delta-sigma modulator specifications as follows: the input signal frequency is 10.7MHz, the sampling frequency is 42.8MHz, the signal bandwidth is 200kHz, oversampling rate is 107, and one bit quantizer.

Due to the demand for high‐performance radio frequency (RF) integrated circuit design in the past years, a system‐on‐chip (SoC) that enables integration of analog and digital parts on the same die has become the trend of the microelectronics industry. As a result, a major requirement of the next generation of wireless devices is to support multiple standards in the same chip‐set. This would enable a single device to support multiple peripheral applications and services. Based on the aforementioned, the traditional superheterodyne front‐end architecture is not suitable for such applications as it would require a complete receiver for each standard to be supported. A more attractive alternative is the highintermediate frequency (IF) radio architecture. In this case the signal is digitalized at an intermediate frequency such as 200MHz. As a consequence, the baseband operations, such as down‐conversion and channel filtering, become more power and area efficient in the digital domain. Such architecture releases the specifications for most of the front‐end building blocks, but the linearity and dynamic range of the ADC become the bottlenecks in this system. The requirements of large bandwidth, high frequency and enough resolution make such ADC very difficult to realize. Many ADC architectures were analyzed and Continuous‐Time Bandpass Sigma‐Delta (CT‐BP‐ΣΔ) architecture was found to be the most suitable solution in the high‐IF receiver architecture since they combine oversampling and noise shaping to get fairly high resolution in a limited bandwidth. A major issue in continuous‐time networks is the lack of accuracy due to powervoltage‐ temperature (PVT) tolerances that lead to over 20% pole variations compared to their discrete‐time counterparts. An optimally tuned BP ΣΔ ADC requires correcting for center frequency deviations, excess loop delay, and DAC coefficients. Due to these undesirable effects, a calibration algorithm is necessary to compensate for these variations in order to achieve high SNR requirements as technology shrinks. In this work, a novel linearization technique for a Wideband Low‐Noise Amplifier (LNA) targeted for a frequency range of 3‐7GHz is presented. Post‐layout simulations show NF of 6.3dB, peak S21 of 6.1dB, and peak IIP3 of 21.3dBm, respectively. The power consumption of the LNA is 5.8mA from 2V. Secondly, the design of a CMOS 6th order CT BP‐ΣΔ modulator running at 800 MHz for High‐IF conversion of 10MHz bandwidth signals at 200 MHz is presented. A novel transconductance amplifier has been developed to achieve high linearity and high dynamic range at high frequencies. A 2‐bit quantizer with offset cancellation is alsopresented. The sixth‐order modulator is implemented using 0.18 um TSMC standard analog CMOS technology. Post‐layout simulations in cadence demonstrate that the modulator achieves a SNDR of 78 dB (~13 bit) performance over a 14MHz bandwidth. The modulator’s static power consumption is 107mW from a supply power of ± 0.9V. Finally, a calibration technique for the optimization of the Noise Transfer Function CT BP ΣΔ modulators is presented. The proposed technique employs two test tones applied at the input of the quantizer to evaluate the noise transfer function of the ADC, using the capabilities of the Digital Signal Processing (DSP) platform usually available in mixed‐mode systems. Once the ADC output bit stream is captured, necessary information to generate the control signals to tune the ADC parameters for best Signal‐to‐Quantization Noise Ratio (SQNR) performance is extracted via Least‐ Mean Squared (LMS) software‐based algorithm. Since the two tones are located outside the band of interest, the proposed global calibration approach can be used online with no significant effect on the in‐band content.

The summing amplifier and the quantizer form two of the most critical blocks in a continuous time delta sigma (CT ΔΣ) analog-to-digital converter (ADC). Most of the conventional CT ΔΣ ADC designs incorporate a voltage summing amplifier and a voltage-mode quantizer. The high gain-bandwidth (GBW) requirement of the voltage summing amplifier increases the overall power consumption of the CT ΔΣ ADC. In this work, a novel method of performing the operations of summing and quantization is proposed. A current-mode summing stage is proposed in the place of a voltage summing amplifier. The summed signal, which is available in current domain, is then quantized with a 3-bit current mode flash ADC. This current mode summing approach offers considerable power reduction of about 80% compared to conventional solutions [2]. The total static power consumption of the summing stage and the quantizer is 5.3mW. The circuits were designed in IBM 90nm process. The static and dynamic characteristics of the quantizer are analyzed. The impact of process and temperature variation and mismatch tolerance as well as the impact of jitter, in the presence of an out-of-band blocker signal, on the performance of the quantizer is also studied.

Abstract In this dissertation, person identification systems based on Karhunen-Loeve transform (KLT) are investigated. Both speaker and face recognition are considered in our design. Among many aspects of the system design issues, three important problems: how to improve the correct classification rate, how to reduce the computational cost and how to increase the robustness property of the system, are addressed in this thesis. Improvement of the correct classification rate and reduction of the computational cost for the person identification system can be accomplished by appropriate feature design methodology. KLT and hard-limited KLT (HLKLT) are proposed here to extract class related features. Theoretically, KLT is the optimal transform in minimum mean square error and maximal energy packing sense. The transformed data is totally uncorrelated and it contains most of the classification information in the first few coordinates. Therefore, satisfactory correct classification rate can be achieved by using only the first few KLT derived eigenfeatures. In the above data transformation process, the transformed data is calculated from the inner products of the original samples and the selected eigenvectors. The computation is of course floating point arithmetic. If this linear transformation process can be further reduced to integer arithmetic, the time used for both person feature training and person classification will be greatly reduced. The hard-limiting process (HLKLT) here is used to extract the zero-crossing information in the eigenvectors, which is hypothesized to contain important information that can be used for classification. This kind of feature tremendously simplifies the linear transformation process since the computation is merely integer arithmetic. In this thesis, it is demonstrated that the hard-limited KL transform has much simpler structure than that of the KL transform and it possess approximately the same excellent performances for both speaker identification system and face recognition system. Moreover, a hybrid KLT/GMM speaker identification system is proposed in this thesis to improve classification rate and to save computational time. The increase of the correct rate comes from the fact that two different sets of speech features, one from the KLT features, the other from the MFCC features of the Gaussian mixture speaker model (GMM), are applied in the hybrid system. Furthermore, this hybrid system performs classification in a sequential manner. In the first stage, the relatively faster KLT features are used as the initial candidate selection tool to discard those speakers with larger separability. Then in the second stage, the GMM is utilized as the final speaker recognition means to make the ultimate decision. Therefore, only a small portion of the speakers needed to be discriminated in the time-consuming GMM stage. Our results show that the combination is beneficial to both classification accuracy and computational cost. The above hybrid KLT/GMM design is also applied to a robust speaker identification system. Under both additive white Gaussian noise (AWGN) and car noise environments, it is demonstrated that accuracy improvement and computational saving compared to the conventional GMM model can be achieved. Genetic algorithm (GA) is proposed in this thesis to improve the speaker identification performance of the vector quantizer (VQ) by avoiding typical local minima incurred in the LBG process. The results indicates that this scheme is useful for our application on recognition and practice.

<p>Adaptive Resource Allocating Vector Quantizer (ARAVQ) är en teknik för datareducering för mobila robotar. Tekniken har visats framgångsrik i enkla miljöer och det har spekulerats i att den kan fungera som ett generellt datautvinningsverktyg för tidsserier. I rapporten presenteras experiment där ARAVQ används som datareducerare på en artificiell respektive en fysiologisk datamängd inom en datautvinningskontext. Dessa datamängder skiljer sig från tidigare robotikmiljöer i och med att de beskriver objekt med diffusa eller överlappande gränser i indatarymden. Varje datamängd klassificeras efter datareduceringen med hjälp av artificiella neuronnät. Resultatet från experimenten tyder på att klassificering med ARAVQ som datareducerare uppnår ett betydligt lägre resultat än om ARAVQ inte används som datareducerare. Detta antas delvis bero på den låga generaliserbarheten hos de lösningar som skapas av ARAVQ. I diskussionen föreslås att ARAVQ skall kompletteras med en funktion för grannskap, motsvarande den som finns i Self-Organizing Map. Med ett grannskap behålls relationerna mellan de kluster som ARAVQ skapar, vilket antas minska följderna av att en beskrivning hamnar i ett grannkluster</p>