USRP2 implementation of compressive sensing based channel estimation in OFDM

Radio channel impairment is a major concern in any wireless system, and channel estimation is used to calculate the multipath channel effects. The traditional way of using pilots for channel estimation has a tradeoff between spectral efficiency and estimation accuracy. An increasing amount of research has been done based on a novel signal processing technique called compressive sensing (CS) and its applications in the modern day wireless systems. In this paper, we can exploit the sparsity of the time domain channel by choosing the pilots randomly and building a random projection measurement matrix, thereby conserving the bandwidth. We investigate the implementation and the challenges in an open source software based radio development kit, GNU Radio for a wireless system, and build a CS-based channel estimator for the OFDM module on a Universal Software Radio Peripheral 2 (USRP2). Simulations are conducted to prove the effectiveness over traditional channel estimation. A Python signal processing package is developed and performance studies are conducted in the real-time system through USRP2.