How to unwire wireless networks

Published Apr 19, 2011

An entirely wireless network would be cheaper for the consumer and could be used in areas that lack infrastructure or where there is a severe risk of earthquakes. However, most of today’s wireless networks still remain wired. This is partly because, over the last 40 years, researchers have not been able to determine network capacity. Nicolas Schrammar and Mikael Skoglund have been given an award for shedding light on this problem using an approximation model.

In most modern wireless networks, only the last link of the communication chain is wireless. A cell phone, for example, is connected wirelessly to the base station, or a computer is connected to a wireless router. The rest of the network is, essentially, still wired.

An entirely wireless network without a wired backbone would have many advantages: it would be cheaper, much more flexible and it could be used in situations where there is no infrastructure, such as after an earthquake or in undeveloped areas.

However the question is, could such a network work in reality? And how much data could be sent over an entirely wireless network?

In order to identify the highest possible data rate manageable by a wireless network, researchers in information theory have been struggling to determine capacity by analysing a mathematical model of networks, without making any actual experiments or simulations. For 40 years. However, after all this time the capacity of most networks still remains unknown.

A recent idea is to use an alternative mathematical model that approximates the capacity of networks with the hope that an approximation would be easier to determine, while still providing enough insight.

Nicolas Schrammar, a PhD student at Communication Theory has, together with his supervisor Professor Mikael Skoglund, won the best paper award at the Wireless Communications and Networking Conference (WCNC) 2011 (in the physical layer track). In their paper “Capacity Bounds for the Discrete Superposition Model of the Gaussian Multiple-Access Channel” they bring more clarity to the problem of establishing network capacity.

“In our paper we find a capacity approximation by considering an alternative model, the discrete superposition model. The approximation works in two steps. Firstly, we know that the capacity of our wireless network is approximated by the capacity in the discrete superposition model. Secondly, we approximate this capacity in the alternative model. The result is an approximation of an approximation of the original capacity,” explains Nicolas Schrammar.

Historically researchers have studied the capacity of the point-to-point channel (one transmitter, one receiver), which led to the development of channel codes. This is still the essential backbone of all modern wireless communication. However as the nature of our communications infrastructure has changed, the multiple channel is today’s reality. Nicolas Schrammar and Mikael Skoglund are specifically looking at approximation of multiple-access channels, which is basically fundamental research. Nevertheless, it is an important piece of the puzzle that is how to unwire wireless networks.

“Having an approximation is better than nothing. Hopefully, those approximations can give us insights into what future wireless networks should look like. Which structures – what part should be talking to what part - are most beneficial, and what coding strategies are needed?”

Why is this important?

“Knowledge of actual capacity can be used to design communication networks from a fundamental point of view, in the sense that it can be decided if one design is fundamentally better that another. Hopefully, capacity approximations can be used for the same purpose. Also knowing the capacity of communication networks will probably stimulate the development of channel codes in order to achieve this capacity in real world implementations.”