Iterative Collision Resolution in Wireless Networks

Abstract

With the growing popularity of smart phones and tablets, development of multimedia applications is on the rise. Speedy transmission of this massive amount information is already pushing the limits of the capacity of wireless networks, and in upcoming years wireless data traffic is projected to continue increasing dramatically. Advances in wireless network throughput are necessary to keep up with society’s data demands.

In an uncoordinated wireless communications system, transmissions collide and interfere as multiple users transmit data to a central receiver. Slotted-ALOHA, the conventional method that schedules user transmissions, has only 37% throughput efficiency. However, theoretical results in recent studies suggest that scheduling transmissions over a number of random, fixed timeslots and employing iterative collision resolution techniques achieves optimal throughput efficiency of approximately 100%. This research considers how real-world conditions affect these theoretical results. A MATLAB model was developed create random graphs, representing users transmitting packets over such timeslots, and the packets were resolved by this method. This model was simulated extensively, representing networks of up to 10,000 users over 10,000 to 15,000 timeslots, and the number of packets resolved in each iteration was measured.

These simulations have generated empirical data that backs up the theoretical claim. The distribution of the number of packets resolved also led to modifications of the current model to increase the percentage of packets resolved in each iteration. Furthermore, by investigating low density parity check coding techniques, other modifications to the current model can optimize the throughput over several consecutive transmissions where unresolved packets are resent with higher probability of recovery. These results demonstrate the potential of this method in handling uncoordinated transmissions in communications systems, even in the presence of finite conditions. This suggests that this method could eventually be employed in actual wireless systems.