Traffic Convexity Aware Cellular Networks: A Vehicular Heavy User Perspective

TL;DR

This paper investigates the rise of vehicular heavy user traffic in cellular networks through field experiments, and proposes a convexity-aware cell association algorithm to improve network performance for such users.

Contribution

It introduces a novel user demand pattern based on velocity, and develops a cell association method that exploits this convex traffic pattern to address vehicular heavy user challenges.

Findings

Vehicular heavy user traffic increased 8.62 times from 2012 to 2015.

The proposed convexity-aware association outperforms traditional load balancing methods.

Field experiments confirm the convex-shaped traffic pattern over user velocity.

Abstract

Rampant mobile traffic increase in modern cellular networks is mostly caused by large-sized multimedia contents. Recent advancements in smart devices as well as radio access technologies promote the consumption of bulky content, even for people in moving vehicles, referred to as vehicular heavy users. In this article the emergence of vehicular heavy user traffic is observed by field experiments conducted in 2012 and 2015 in Seoul, Korea. The experiments reveal that such traffic is becoming dominant, captured by the 8.62 times increase in vehicular heavy user traffic while the total traffic increased 3.04 times. To resolve this so-called vehicular heavy user problem (VHP), we propose a cell association algorithm that exploits user demand diversity for different velocities. This user traffic pattern is discovered first by our field trials, which is convex-shaped over velocity, i.e. walking user traffic is less than stationary or vehicular user traffic. As the VHP becomes severe, numerical evaluation verifies the proposed user convexity aware association outperforms a well-known load balancing association in practice, cell range expansion (CRE). In addition to the cell association, several complementary techniques are suggested in line with the technical trend toward 5G.