DSP implementation of SNR estimation algorithm for OFDM systems

The objective of this paper is to present the implementation and optimization process of a Signal-to-Noise Ratio (SNR) estimation algorithm operating on BPSK modulated data targeted to multicarrier systems in Additive White Gaussian Noise (AWGN) channels on the C6711 Digital Signal Processor (DSP) of Texas Instruments (TI). In our case, the algorithm was applied to an Orthogonal Frequency Division Multiplexing (OFDM) system. First, the basic structure of the algorithm was implemented and tested using plain C language and Matlab-generated OFDM samples for proof of concept reasons. Next, the code was ported on the DSP for implementation and optimization using DSP-specific C/Assembly, and embodied into a previously implemented, in our lab, OFDM transceiver running on the DSP. The optimization scheme followed was based on several optimization aspects, such as identifying and recoding the most time-consuming parts utilizing the target hardware's characteristics, as well as algorithmic modifications and compiler features. The correctness of the optimization was continuously verified at every optimization stage by providing to the estimator actual data having passed through the transceiver. Finally, a comparison of the implemented estimator with another implemented SNR estimator, in terms of execution speed and SNR estimation accuracy, is presented.