Mobile Data Offloading through A Third-Party WiFi Access Point: An Operator's Perspective

TL;DR

This paper investigates mobile data offloading to third-party WiFi access points from an operator's perspective, analyzing different interference cancellation scenarios and proposing a distributed offloading scheme to maximize utility.

Contribution

It formulates the offloading problem under various SIC scenarios, proves near-optimality and structural properties of solutions, and introduces a practical distributed offloading scheme.

Findings

SIC decoders improve operator utility.

Optimal user association varies with SIC availability.

Distributed scheme performs well with proper threshold tuning.

Abstract

WiFi offloading is regarded as one of the most promising techniques to deal with the explosive data increase in cellular networks due to its high data transmission rate and low requirement on devices. In this paper, we investigate the mobile data offloading problem through a third-party WiFi access point (AP) for a cellular mobile system. From the cellular operator's perspective, by assuming a usage-based charging model, we formulate the problem as a utility maximization problem. In particular, we consider three scenarios: (i) successive interference cancellation (SIC) available at both the base station (BS) and the AP; (ii) SIC available at neither the BS nor the AP; (iii) SIC available at only the BS. For (i), we show that the utility maximization problem can be solved by considering its relaxation problem, and we prove that our proposed data offloading scheme is near-optimal when the number of users is large. For (ii), we prove that with high probability the optimal solution is One-One-Association, i.e., one user connects to the BS and one user connects to the AP. For (iii), we show that with high probability there is at most one user connecting to the AP, and all the other users connect to the BS. By comparing these three scenarios, we prove that SIC decoders help the cellular operator maximize its utility. To relieve the computational burden of the BS, we propose a threshold-based distributed data offloading scheme. We show that the proposed distributed scheme performs well if the threshold is properly chosen.