http://hdl.handle.net/1721.1/4059 http://dspace.mit.edu:80/retrieve/4141 http://hdl.handle.net/1721.1/4059 http://hdl.handle.net/1721.1/46716 Scaling Laws for Heterogeneous Wireless Networks Niesen, Urs This thesis studies the problem of determining achievable rates in heterogeneous wireless networks. We analyze the impact of location, traffic, and service heterogeneity. Consider a wireless network with n nodes located in a square area of size n communicating with each other over Gaussian fading channels. Location heterogeneity is modeled by allowing the nodes in the wireless network to be deployed in an arbitrary manner on the square area instead of the usual random uniform node placement. For traffic heterogeneity, we analyze the n × n dimensional unicast capacity region. For service heterogeneity, we consider the impact of multicasting and caching. This gives rise to the n × 2n dimensional multicast capacity region and the 2n × n dimensional caching capacity region. In each of these cases, we obtain an explicit informationtheoretic characterization of the scaling of achievable rates by providing a converse and a matching (in the scaling sense) communication architecture. Thesis Supervisor: Devavrat Shah Title: Associate Professor Thesis Supervisor: Gregory W. Wornell Title: Professor http://hdl.handle.net/1721.1/46306 Collision Helps! An Analytical Study of ZigZag Decoding Gheibi, Ali Parandeh Sundararajan, Jay Kumar M´edard, Muriel http://hdl.handle.net/1721.1/43953 Estimation and Calibration Algorithms for Distributed Sampling Systems Divi, Vijay Traditionally, the sampling of a signal is performed using a single component such as an analog-to-digital converter. However, many new technologies are motivating the use of multiple sampling components to capture a signal. In some cases such as sensor networks, multiple components are naturally found in the physical layout; while in other cases like time-interleaved analog-to-digital converters, additional components are added to increase the sampling rate. Although distributing the sampling load across multiple channels can provide large benefits in terms of speed, power, and resolution, a variety mismatch errors arise that require calibration in order to prevent a degradation in system performance. In this thesis, we develop low-complexity, blind algorithms for the calibration of distributed sampling systems. In particular, we focus on recovery from timing skews that cause deviations from uniform timing. Methods for bandlimited input reconstruction from nonuniform recurrent samples are presented for both the small-mismatch and the low-SNR domains. Alternate iterative reconstruction methods are developed to give insight into the geometry of the problem. From these reconstruction methods, we develop time-skew estimation algorithms that have high performance and low complexity even for large numbers of components. We also extend these algorithms to compensate for gain mismatch between sampling components. To understand the feasibility of implementation, analysis is also presented for a sequential implementation of the estimation algorithm. In distributed sampling systems, the minimum input reconstruction error is dependent upon the number of sampling components as well as the sample times of the components. We develop bounds on the expected reconstruction error when the time-skews are distributed uniformly. Performance is compared to systems where input measurements are made via projections onto random bases, an alternative to the sinc basis of time-domain sampling. From these results, we provide a framework on which to compare the effectiveness of any calibration algorithm. Finally, we address the topic of extreme oversampling, which pertains to systems with large amounts of oversampling due to redundant sampling components. Calibration algorithms are developed for ordering the components and for estimating the input from ordered components. The algorithms exploit the extra samples in the system to increase estimation performance and decrease computational complexity. Thesis Supervisor: Gregory W. Wornell Title: Professor of Electrical Engineering and Computer Science http://hdl.handle.net/1721.1/41931 A Comparison of Parallel Gaussian Elimination Solvers for the Computation of Electrochemical Battery Models on the Cell Processor Geraci, James R. The rising cost of fossil fuels, together with a push for more eco-friendly methods of transportation, has increased interest in and demand for electrically powered or assisted vehicles. The majority of these electric or hybrid electric vehicles will be, for the foreseeable future, powered by batteries. One of the major problems with batteries is their aging. For batteries, aging means that the maximum charge they can store decreases as number of charge/discharge cycles increases. Aging also means that after a certain number of charge/discharge cycles, the battery will fail. In lead-acid batteries, one of the major phenomenon that promotes battery failure is the development of a non-uniform concentration gradient of electrolyte along the electrodes’ height. This phenomenon is known as electrolyte stratification. This thesis develops a simple two-level circuit model that can be used to model electrolyte stratification. The two-level circuit model is justified experimentally using digital Mach-Zehnder interferometry and is explained theoretically by means of two different electrochemical battery models. The experiments show how the usage of the electrode varies along its height while the simulations indicate that the high 3 resistivity of the lead dioxide electrode plays a major role in the development of a stratified electrolyte. Finally, computational issues associated with the computation of a sophisticated two dimensional electrochemical battery model on the multicore Cell Broadband Engine processor are addressed in detail. In particular, three different banded parallel Gaussian elimination solvers are developed and compared. These three solvers vividly illustrate how performance achieved on the new multicore processors is strongly dependent on the algorithm used. Thesis Supervisor: John L. Wyatt, Jr. Title: Professor Thesis Supervisor: Thomas A. Keim Title: Principal Research Engineer