Conflict Resolution in Multiple Access Channels Supporting Simultaneous Successful Transmissions

TL;DR

This paper introduces a new conflict resolution algorithm for multiple access channels that efficiently handles simultaneous transmissions, achieving near-optimal performance proportional to the number of messages that can be transmitted concurrently.

Contribution

It presents a novel non-adaptive conflict resolution algorithm with a proven ratio of 1/d compared to the optimal, along with new combinatorial structures and bounds.

Findings

The algorithm's time slots scale linearly with d, the number of concurrent messages.

A lower bound within a log(k/d) factor of the upper bound is established.

New combinatorial codes and selector generalizations are introduced for conflict resolution.

Abstract

We consider the Conflict Resolution Problem in the context of a multiple-access system in which several stations can transmit their messages simultaneously to the channel. We assume that there are n stations and that at most k<= n stations are active at the same time, i.e, are willing to transmit a message. If in a certain instant at most d<=k active stations transmit to the channel then their messages are successfully transmitted, whereas if more than d active stations transmit simultaneously then their messages are lost. In this latter case we say that a conflict occurs. The present paper investigates non-adaptive conflict resolution algorithms working under the assumption that active stations receive a feedback from the channel that informs them on whether their messages have been successfully transmitted. If a station becomes aware that its message has been correctly sent over the channel then it becomes immediately inactive. The measure to optimize is the number of time slots needed to solve conflicts among all active stations. The fundamental question is whether this measure decreases linearly with the number d of messages that can be simultaneously transmitted with success. We give a positive answer to this question by providing a conflict resolution algorithm that uses a 1/d ratio of the number of time slots used by the optimal conflict resolution algorithm for the case d=1. Moreover, we derive a lower bound on the number of time slots needed to solve conflicts non-adaptively which is within a log (k/d) factor from the upper bound. To this aim, we introduce a new combinatorial structure that consists in a generalization of Komlos and Greenberg codes. Constructions of these new codes are obtained via a new generalization of selectors, whereas the non-existential result is implied by a non-existential result for a new generalization of the locally thin families.