Software Defined Radio Platform for Cross RF/PHY/Network/Application Layer Study in Cloud Random Access Networks

点击翻译按钮获取中文摘要

By 2016, mobile connection speeds will increase 9-fold, aggregated smartphone traffic will be 50 times greater than it is today, and mobile video will increase 25-fold [Cisco Global Mobile Data Traffic Forecast]. In response, wireless networks have been constantly evolving to support higher data rates and wider coverage, at a lower cost. The bottleneck in wireless connectivity is the spectrum, a scarce natural resource that in addition to its rising price, has a hard limit imposed by nature. As an example at the national level, in January 2014, an auction for 68 MHz of Ultra High Frequency (UHF) spectrum was held in Canada, generating a total value of $5.27b. The problem is that, regardless of the dollar amount, the entire usable spectrum is already exploited and in most parts is overcrowded. It is clear at this stage of the wireless evolution that the only viable option is based on using smaller cells, which act cooperatively and realize a frequency reuse of one without causing too much interference to each other. To keep up with the cost of implementing many such small cells, 'Cloud Radio Access Networks (CRAN)' are pursued as a mechanism to share resources. In conventional homogeneous networks, each base-station directly serves the clients in its neighborhood. To increase capacity and coverage, future wireless networks will have a heterogeneous architecture, in which a high power macro base-station along with some low-power nodes across its cell, in particular distributed-antenna-systems with remote radio heads, collectively serve clients. This is the path wireless networks will follow for many years to come. However, to realize cost effective solution, a number of issues need to be addressed. One of the main bottlenecks is the connection of radio head to central node. The only possible way to study the actual tradeoffs in the implementation of CRAN is a realistic platform implemented on top of a loaded network of real-life users. The requested programmable network infrastructure, installed on top of the uWaterloo network, will provide a realistic environment to study and optimize CRAN performance under realistic loads. This in turn will allow avoiding unexpected surprises after costly deployments.