March 2, 2017

On user and traffic distribution for extreme rural scenario

Radio Access Network, November 2016

By: Ali Panah

Abstract

A fully connected society is expected in the near future resulting in a tremendous growth in connectivity, traffic volume and diversification of usage scenarios [1]. This generates the need to improve system efficiencies in terms of spectrum, energy, operation, etc. To better serve the needs of such a fully connected and networked society in the near future, in both developing and developed countries, TR 38.913 states typical usage scenarios and attributes associated with them [2]. The family of use cases include eMBB (enhanced Mobile Broadband), mMTC (massive Machine Type Communications) and URLLC (Ultra-Reliable and Low Latency Communications).

The main focus of this contribution is on scenario 6.1.6 titled “Extreme long distance coverage in low density areas”. This scenario is characterized by long range (~100 km) macro cells which target low user density regions. Next generation technologies associated with this usage scenario is likely to result in a growth in the number of connected devices and consequently data traffic. To fully characterize this scenario, models related to user and traffic distribution are extremely important.

To this end, experimental studies and user distribution models have been proposed recently. For instance, [3] presents a statistical analysis of measurement data from a live 3G network and [4] proposes an empirical user distribution model based on these measurements. Specifically, it proposes that a uniform distribution assumption be replaced with a log normal distance distribution. While the model is important for service providers to gain an intuition into the traffic characteristics of the underlying network, it should be noted that this model is made based on the activity of users already connected to an existing network. The extrapolation of usage data of currently connected people to the spatial distribution of both connected and unconnected populations might not be completely accurate in practical scenarios. This highlights the need to evaluate and further tune these models.

In this contribution, we leverage new research results and data [5] – population density data (hereafter, popdens) – that provide a high-resolution view into global population density distributions. (These results and data were obtained through the application of state-of-the-art computer vision and machine learning technologies on high resolution satellite imagery combined with existing census data. This massive data set covers over 20 countries in the world).

We conduct a rigorous user distribution model case study that focusses on a specific region. We use a subset of popdens to evaluate the accuracy of previously proposed models and specifications. We use demographic data that provides population distribution over a couple of regions whose size is roughly greater than the size that will be covered by future macro cells associated with scenario 6.1.6. We analyse this data to understand the traffic distribution associated with this region. Further, we compare the results of our analysis with those obtained from the distribution model proposed in [4].