Combined Nyquist and Compressed Sampling Method for Radio Wave Data Compression of a Heterogeneous Network System

Doohwan LEE
Takayuki YAMADA
Hiroyuki SHIBA
Kazuhiro UEHARA

IEICE TRANSACTIONS on Communications   Vol.E93-B    No.12    pp.3238-3247
Publication Date: 2010/12/01
Online ISSN: 1745-1345
DOI: 10.1587/transcom.E93.B.3238
Print ISSN: 0916-8516
Type of Manuscript: Special Section PAPER (Special Section on Wireless Distributed Networks)
compressed sensing,  compressive sampling,  heterogeneous network system,  l1-minimization,  cognitive radio,  

Full Text: FreePDF

To satisfy the requirement of a unified platform which can flexibly deal with various wireless radio systems, we proposed and implemented a heterogeneous network system composed of distributed flexible access points and a protocol-free signal processing unit. Distributed flexible access points are remote RF devices which perform the reception of multiple types of radio wave data and transfer the received data to the protocol-free signal processing unit through wired access network. The protocol-free signal processing unit performs multiple types of signal analysis by software. To realize a highly flexible and efficient radio wave data reception and transfer, we employ the recently developed compressed sensing technology. Moreover, we propose a combined Nyquist and compressed sampling method for the decoding signals to be sampled at the Nyquist rate and for the sensing signals to be sampled at the compressed rate. For this purpose, the decoding signals and the sensing signals are converted into the intermediate band frequency (IF) and mixed. In the IF band, the decoding signals are set at lower center frequencies than those of the sensing signals. The down converted signals are sampled at the rate of four times of the whole bandwidth of the decoding signals plus two times of the whole bandwidth of the sensing signals. The purpose of above setting is to simultaneously conduct Nyquist rate and compressed rate sampling in a single ADC. Then, all of odd (or even) samples are preserved and some of even (or odd) samples are randomly discarded. This method reduces the data transfer burden in dealing with the sensing signals while guaranteeing the realization of Nyquist-rate decoding performance. Simulation and experiment results validate the efficiency of the proposed method.