Internet Performance Observatory
Measuring and Understanding Global Network Connectivity
We need a better Internet. We need a faster Internet for the next generation multimedia services. We need a more reliable Internet to be able to support our healthcare and an older population. We need a more secure Internet where our personal data stays confidential and our national infrastructure is immune to crime and cyber-terrorism.
To achieve this goal we need to better understand global Internet performance. We need to know how much speed users can get, what factors control this and how reliable their connectivity is. We also need to analyse the Internet’s energy consumption to help building greener networks. This is the mission of UCL’s Internet Performance Observatory. Our work is divided in three main tasks:
Measure |
Analyse |
Disseminate |
|||||||||||||||||||
|
We are creating network measurement tools that will measure Internet performance from millions of users in every country in the planet. These probes will be placed in networks all around the world and send the results to our data centre. This data will be added and combined with other datasets from other researchers |
|
|
|
|
|
|
|
|
|
We will be analysing this data in real-time and extract useful knowledge from it. We will use a myriad of Big Data techniques including regression analysis, evolutionary algorithms, machine learning, data mining and deep learning. This will allow us to determine all the factors determining performance, forecast the evolution of the global Internet and detect global and local network problems in real-time |
|
|
|
|
|
|
|
|
|
|
We will disseminate this knowledge both in real-time and through periodic reports on the state of the network. Real-time dissemination will consist of a stream based publish/subscribe system where anybody will be able to obtain information about specific metrics and/or regions of the Internet. Periodic reports will be printed publications with detailed results. |
We belong the UCL Institute for Communications and Connected Systems and are sponsored by a variety of industrial partners. We guarantee full independence in all data collection and analysis.
