Graduate Exam Abstract
May 12, 2010, 12:30pm-2:30pm
ECE conference room
Ph.D. Preliminary Exam
Abstract: Rate and Error Performance of Network Communications with Complexity Constraints. Rate and error probability are two fundamental parameters in a communication system, reflecting the communication efficiency and reliability. Complexity constraint is also an important concern in practical system. The tradeoff of rate and error performance of communication system under various complexity constraints is therefore of great interests and has been intensively studied in classical information theory since Shannon's celebrated work in 1948.
The evolution of point to point communication to network communication has given birth to some new research problems. While rate, error and complexity tradeoff of these problems is still of vital importance, due to the invalidity of some key information theoretic assumptions, classical analysis no long applies directly. Our research aims at extending information theory to some of these network communications under various conditions. This report covers two specific cases.
In the first case, we consider the communication over unknown channels. The challenge is how the transmitter, without knowledge of channel conditions, should adapt to unknown channel changes and to heterogeneous users. This problem can be solved by a new fountain communication model, the basic idea of which is to shift communication rate determination from the transmitter to the receiver. In this report, a new fountain channel coding scheme is developed to achieve near optimal rate and error probability tradeoff with linear encoding/decoding complexity.
In the second case, we consider the communication where transmitters send messages over a common channel distributively, without sharing rate information with each other or the receiver. The receiver chooses to report an estimate of the transmitted messages or claim a collision, depending on whether reliable message recovery is possible. The relaxation of joint communication rate determination among transmitters provides no guarantee of reliable communication. However, fundamental performance of the communication system can still be characterized using an achievable rate region in a sense made clear in the report. The corresponding rate and error probability tradeoff is obtained using large deviation analysis.
The achivability of optimal rate and error performance tradeoff under linear coding complexity in a classical communication system is also obtained.
Adviser: J. Rockey Luo
Non-ECE Member: Anton Betten (Mathematics Dept.)
Member 3: Louis Scharf
Addional Members: N/A
Program of Study: