PACKET COLLISION RESOLUTION IN RANDOM ACCESS MEDIUM USING SEMI-BLIND SOURCE SEPARATION TECHNIQUE

Abstract

In the broadcast networks or random access networks, the main problem encountered is the collision of packets of one user or node with packets of other users or nodes. This leads to irregular or varying delays and throughput is deteriorated. Here signal processing approach has been taken to resolve the collisions. Among the three signal. processing approaches such. as training, blind and semi-blind packet collision resolution techniques, training based methods are inefficient in bandwidth utilization, blind methods are complex. So we chose semi-blind packet separation technique to resolve the collisions. Packet collisions are modeled as multiple input multiple output system with additive white Gaussian noise added. M-ary phase shift keying scheme was used as modulation technique. Semi-blind least squares smoothing with successive subtraction is been used to resolve the collisions. The least squares smoothing has been implemented using oblique projections. RMSE [root mean squared error] between original symbols and estimated symbols of the packet is the performance measure being evaluated. This measure has been evaluated for varying packet sizes, varying signal to noise ratios, varying output number, and M-ary PSK with M varying from 2 to 64, with varying difference in voltage levels assigned to the symbols. The simulation was carried out in MATLAB environment