Channel-aware energy optimization of OFDM receivers using dynamic precision scaling in FPGAS

To reduce the energy consumption of Orthogonal Frequency-Division Multiplexing (OFDM) systems, a new variable word-length method is presented in this paper. A simulation based approach is used: the optimized fixed-point implementation of an OFDM receiver is found for different simulated channel conditions, depending on the Signal-to-Noise Ratio (SNR) and the channel type. During the execution, the receiver estimates the channel conditions and chooses the optimum word-length to decode the received information. A realistic energy consumption of the receiver is estimated with a library that contains the energy consumption of Field-Programmable Gate Array (FPGA) basic operators depending on the bit-width, obtained from experimental data. Up to 57% of the dynamic energy can be saved using this method.