• home
• current students
• graduate
• events

Presenter: Abolfazl Ghassemi
Supervisor:

Date: Tue, July 29, 2003
Time: 10:00:00 - 11:00:00
Place: EOW 430

ABSTRACT

Abstract:

Due to the superior performance of turbo codes, they are employed in future generation wireless communication standards. The use of newer, more powerful error correction codes such as turbo codes accompanies the development of software-defined radio (SDR) systems. In these systems, the accumulation of errors due to fixed-point implementation contributes to the degradation in performance of turbo codes. This degradation can be reduced by increasing the resolution of the analog-to-digital converters (ADC), but at the cost of increasing hardware complexity. With optimum gain control of the signal prior to quantization, it is possible to reduce much of this degradation. However, controlling the signal level in a fading channel is difficult. Decoding algorithms for turbo decoding are iterative in nature. Each decoding iteration results in additional computations and decoding delay. The maximum a posterior (MAP) algorithm is optimal but complex for implementation. Three to eight bit quantization of the signal is examined in order to investigate the feasibility of adapting the resolution ADC in the turbo decoders. An efficient criterion to stop the iteration process and prevent unnecessary computations and decoding latency is obtained using this criterion. In addition, low complexity decoding algorithms similar to the soft output Viterbi algorithm (SOVA) as well as simplified variations of the log-MAP algorithm are investigated. The tradeoffs between reliable backward metrics, memory units and hardware complexity are explored for the full string and sliding window techniques.

Questions?

Please contact by email: aghassem@ece.uvic.ca