Objective
Research objectives and content
Digital communication systems are progressively overtaking their analogue ancestors, due to the great flexibility of information transfer afforded by digital technology. This proposal addresses a key constituent required of autonomous digital receivers: adaptive blind equalization. Blind methods afford channel equalization possibilities without squandering precious channel capacity. In Europe in particular, gains from blind equalization are anticipated in satellite transmission, future generation mobile telephony requiring advanced data transmission rates, and ultimately commercial broadcasting.
Existing blind adaptive equalization algorithms have been employed with encouraging success under special conditions, involving finite-length channel impulse responses, negligible channel noise, and a certain 'channel disparity' condition. However, practical channels often violate these conditions, and this research aims to develop improved adaptive algorithms enjoying robustness with respect to channel undermodelling and parasitic channel interference.
The research work will focus on signal enhancement and separation in convolutional mixtures, refinement of objective functions to avoid undesirable local minima, and projection techniques which greatly accelerate convergence compared to conventional gradient descent implementations.
Training content (objective, benefit and expected impact)
As existing blind equalization schemes show clear performance limitations in realistic signal and channel environments, the development of robust adaptive equalization methods proposed here will be of major importance towards enabling Europe to play a dominant role in future-generation communication technologies. The proposed research will refine expertise in the following areas: a) A better understanding of the factors which limit performance in present-generation equalization algorithms; b) The development of new adaptive algorithms enjoying robustness with respect to channel undermodeling and parasitic channel interference; c) Convergence acceleration techniques which combat the otherwise slow convergence exhibited by present generation algorithms; d) Validation of the developed techniques on actual communication channels, in order to validate improvements in performance and reliable equalization under wide-ranging conditions.
Links with industry / industrial relevance (22)
Fields of science
Topic(s)
Call for proposal
Data not availableFunding Scheme
RGI - Research grants (individual fellowships)Coordinator
91011 Evry
France