Transporting JPEG Images over MIMO Links in Ad-Hoc Networks: an Investigation

This paper reports an investigation into the transport of sensor-generated still images over, ad-hoc, energy- limited networks. MIMO technology was considered as a means of increasing the data rate over network links and thereby the image throughput of the network as a whole. The combining effect of MIMO block size, packet size, and the JPEG compression technique on the distortion of the received images was of particular interest. In addition, we consider two forms of receiving mechanism: maximum likelihood (ML) and the less complex zero-forcing (ZF). We centered our investigation on simulating link behavior starting with a base of MIMO models written in LabView that were previously developed at the University of Texas. Additions and modifications made to this base, verifying calculations written in Matlab, and the utility of Labview and Matlab in the investigation are described. JPEG images were transported over the resulting Labview simulation through Rayleigh fading and AWGN over a range of signal-to- noise ratios, with SISO, 2times2 MIMO, QPSK modulation, ML, and ZF receivers. The resulting bit error rate vs. SNR curves are presented and are in agreement with theory. In addition the resulting JPEG image distortion is described.