FatVAP: Aggregating AP Backhaul Bandwidth
overview - What is Fat-VAP??
papers - Fat-VAP's documents
Related Projects - Press Articles
people - who are we?
funding - who sponsors FatVAP?

Overview

FatVAP Architecture
FatVAP's reverse-NAT like architecture

Increasingly, computers in coffee shops, residential areas and hotspot scenarios see multiple open access points (APs). For e.g., cafe's and restaurants provide free internet, cities provide pole-top municipal networks and many residential users open up their access points.

This creates an interesting scenario. The wireless link to these APs is often high speed, 30Mbps achievable with 802.11a and even more with the newer 802.11n. Yet, the net bandwidth one can get through the AP is bottlenecked by relatively low throughput DSL or cable modem links that connect the APs to the Internet.

Ideally, a client would want to simultaneously use all accessible APs and obtain the sum of their backhaul bandwidth. Past work can connect to multiple APs, but can neither maintain concurrent TCP connections across them, nor can it aggregate AP backhaul bandwidth.

FatVAP is an 802.11 driver that aggregates the bandwidth available at accessible APs and also balances user load across the APs. FatVAP has three key features. First, it knows exactly how long to connect to each AP in order to collect the bandwidth it can ever get from this AP. Second, FatVAP switches quickly between APs and without losing queued packets, making it the only driver that can sustain concurrent high throughput TCP connections across multiple APs. Third, FatVAP works with unmodified APs and is transparent to applications and the rest of the network stack.

Experiments with FatVAP both in our lab and actual residential deployments. Our results show that, in today's deployments, FatVAP allows improves user's bittorrent throughput by upto 2.6 times and reduces his firefox's response time to download web pages by upto 2.8 times.

Papers

Related Projects

Software

People


NMS@MIT CSAIL
M. I. T. Computer Science and Artificial Intelligence Laboratory The Stata Center, 32 Vassar St. Cambridge, MA 02139 USA