NSF Award
1936572
Internet traffic traveling over wireless networks, such as 5G and wireless mesh networks, inevitably experiences loss, i.e., a fraction of sent data packets are lost and never arrive at their destination. Loss is exacerbated when senders or receivers are mobile. The current design of the internet inefficiently deals with loss by workarounds such as retransmitting lost data. This project envisions a redesign of the internet that will accommodate loss by fundamentally reimagining how internet data is stored and transmitted, as well as an investigation of smaller incremental steps that might be taken to facilitate this larger redesign. Improving the capability of wireless networks in this way will enable deployment of a more reliable, robust, responsive, and efficient internet to underserved and rural communities around the world that rely on wireless connectivity.<br/><br/>The proposal is to flesh out an internet design based on a fountain code, i.e., all data stored or transmitted within the internet will be fountain encoded data. A fountain code is an erasure code that is scalable in two dimensions: (1) the size of an object which can be efficiently encoded and decoded is essentially unlimited; and (2) the amount of encoded data that can be generated from an object is essentially unlimited. A fountain code has (nearly) optimal recovery, i.e., any portion of encoded data (nearly) equal in size to an object is sufficient to recover the object. The design will separate loss recovery from congestion control, incorporate universal mechanisms to simply and gracefully handle packet loss, mobility, long latency paths, and failing infrastructure, include mechanisms to address novel security concerns, and take into account deployment concerns.<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
