T. Aaron Gulliver
Project Topics
1.
Turbo Codes
Description: Turbo codes provide a means of achieving near Shannon capacity performance with a reasonable amount of decoding complexity. The key to this performance is an iterative soft output decoding algorithm. To date, turbo codes are not well understood, particularly the role of the random interleaver and the structures of the component codes. This project will explore this class of codes (as well as classes of random-like codes) with the aim of developing decoding algorithms and codes with good performance/complexity tradeoffs. In particular, simple low complexity codes will be considered as component codes because of their simple structures. The graph structure of these codes is also of interest. The graph structure of these codes is also of interest.

2.
Neural Network Diversity Combining for Fast Frequency Hopped Communications
Description: Symbol repetition is used in frequency hopped spread spectrum communication systems as a simple means of providing protection against interference and fading. Many of the techniques used for combining these symbols have limitations which can cause system failures in certain types of channel conditions. This project will consider the development of new combining techniques based on a neural network. The goal is to provide robustness to changing interference patterns to ensure the greatest system reliability and performance.

3.
Trellis Structures for Linear Block Codes
Description: Convolutional codes have traditionally been favored over block codes for communication systems because soft decision decoding is possible using a trellis (Viterbi) decoder. As well, trellis coded modulation has successfully combined convolutional coding and modulation. However, convolutional codes are not well suited to packet type communication systems, among others. The soft decision decoding of block codes has recently been characterized, but suffers from the structure of existing block codes which were not developed for trellis decoding. This project considers the design of block codes specifically for decoding using trellis decoders. The theory will be developed to systematically construct codes which have simple trellis structures. Block trellis coded modulation is a related problem which can also be investigated. Bounds can also be examined to determine the achievable coding gains.

4.
Spread Spectrum Multimedia Communications
Description: The tremendous increases in demand for personal and mobile communications has created a requirement for integrated services such as voice, data, images and video (often using ATM). Since different users require different quality of service (QOS), the channel resources should not be shared equally among the various classes of users. The multiple access capability of the CDMA channel can be used to accommodate voice calls (and other real-time services), while the data users follow the ALOHA protocol. One means of achieving control over the QOS is through the error control coding mechanisms that are employed. This project will consider adaptive coding techniques as a means of improving capacity while maintaining QOS.

5.
Design of Good Linear Error Correcting Codes
Description: One of the fundamental problems in algebraic coding theory is the determination of bounds on the maximum achievable distance between codewords for a given code size. Lower bounds are usually devised through construction techniques which provide a code with the required parameters. A good code is defined as one which achieves the largest known minimum distance. This project will investigate the development of techniques and/or application of existing techniques for constructing good codes which improve these bounds. In particular, combinatorial heuristics will be considered.

6.
Multiuser Communications with Nonorthogonal Signalling
Description: A large number of multiuser communications systems rely on orthogonal signalling. Although this is effective, it is not a criterion of optimality. This project will consider signal design based on maximizing the minimum Euclidean distance in the overall summed set of signals.

7.
Coding for Error Control in Computer Memory Systems
Description: The density of modern computer memory components makes them susceptible to interference and failure. Single event upsets are soft errors (and sometimes hard) caused by atomic particles travelling through the device, and normally appear as random bit errors. Component failures and damage to bus lines lead to hard byte type errors, depending on the memory configuration. As well, adjacent errors commonly occur due to the device structure. Majority logic decodable codes which can correct random and adjacent errors are currently being considered because of the ease and speed of decoding. Future research will consider the design of codes which can correct both bit errors and byte errors/erasures to overcome both random bit errors and block (byte) memory failures. In addition, error detection in conjunction with error correction will be investigated.

8.
Self-Dual Codes and Weighing Matrices
Description: Weighing matrices provide a rich source of constructions for self-dual or orthogonal codes. However, few constructions are known to provide extremal, or optimal codes. This project will investigate the theory of weighing matrices with the goal of defining extremal self-dual codes.

9.
Design of Fast and Secure Software Cryptographic Systems
Cryptography has become an important part of many digital communications systems. Because of the ever increasing use of the internet for commerce, improved techniques for the electronic transfer of funds is required. The algorithms and their implementations must be simple and fast. This project will look at developing new software based cryptographic systems that will achieve these goals.

10.
Chaotic Cryptographic Systems
Chaotic systems show great promise for use in cryptographic systems. The problems left to be solved are how to integrate these systems with digital data, and also the reproducibility of the system outputs. This project will explore these issues as well as the cryptographic strength of such systems.

11.
Coding for Frequency-Hopped MFSK
Error control coding is an essential part of every spread spectrum communication system. This project will consider the use of Reed-Solomon and related burst error correcting codes in a frequency hopped system. The goal is to obtain a robust and efficient solution in fading and multiple-access interference.

12.
Digital Watermarking
In the manufacturer of paper, wet fiber is subjected to high pressure to expel the moisture. If the mold has a slight pattern, this pattern leaves an imprint in the paper, called a watermark. Digital watermarks are imperceptible, or barely perceptible, transformations of digital data; often the digital data set is a digital multimedia object. Watermarks can be applied to almost any form of digital data, for example, videos and music. This project will consider the development of a digital watermarking system for a particular class of media.


Aaron Gulliver
2000-02-21