the project
Project Techinical Considerations

Our methodology employs a combination of theory, algorithm and protocol design, simulation, implementation, and testbed experimentation. To realize our driving vision of a scalable, deployable, high-performance Internet, we will make three distinct fundamental technical contributions addressing multiple timescales of control (see Figure 2) that will be integrated into our testbed for prototype deployment.

Figure 2: System Timescales

Capacity Scaling via Distributed Opportunistic Scheduling and Medium Access

Fundamental to the TAP network, scheduling and MAC protocols have to be distributed (scheduling decisions cannot be centralized), multi-channel (TAPs can simultaneously send/receive multiple packets via MIMO), and multi-destination (packets originating from MUs are targeted for any wired TAP rather than a fixed destination). We will design, analyze, and build an opportunistic scheduler and contention-minimizing MAC that will aim to follow a hypothetical global schedule in a distributed manner. In opportunistically selecting the best node, channel, or destination based on current channel conditions, we will explicitly address scalability and provide foundations for designing high-performance scheduling and MAC protocols for large-scale wireless networks.

Coordinated Resource Management

To achieve deployability and system-wide high performance, the TAP network must mitigate, or ideally eliminate, spatial bias of throughput to ensure that flows closest to wired TAPs do not receive a disproportionately greater share of bandwidth than nodes multiple hops away. Moreover, to achieve scalability, the TAP network must maximally exploit spatial reuse. To achieve these goals, we will design a coordinated resource management algorithm that ensures that flows are throttled to their fair rate at their network ingress point. Our methodology will balance the need to achieve fairness with more aggressive forwarding that ensures that a sufficient number of packets are backlogged at TAPs to exploit opportunistic medium access when high quality channels permit, or when contention and congestion are temporarily reduced.

The Network is the Channel - Estimation, Routing and Capacity Scaling

Information-theoretic analyses of the capacity of wireless networks have taken an overly optimistic view by ignoring the critical performance impacts of protocols such as medium access and routing. With a unique viewpoint of treating the "whole network as a channel," we will perform a scaling analysis of the TAP network and wireless networks in general that characterizes crucial tradeoffs between protocols and network capacity. With this analytical framework, we will design, implement, and analyze a highly scalable routing protocol that integrates on-demand reactive routing among MUs with periodic proactive routing within the TAP backbone.

To realize this vision, we will implement proof-of-concept prototype TAPs and enhanced IEEE 802.11- compliant MUs and develop a testbed that includes multiple TAPs on the Rice campus, residences in surrounding neighborhoods, and nearby business areas. In particular, we will develop custom hardware based on Verilog/VHDL programmed FPGAs implementing the above media access and scheduling algorithms, integrated with a 2.4 GHz MIMO radio chipset that is capable of providing fast timescale channel measurements to the MAC. This implementation will provide a first-of-its-kind platform for demonstrating the scalability of the TAP architecture and algorithms and will provide critical insights and measurements for understanding the future high-speed wireless Internet.

Home  |  The Project  |  Architecture  |  Research  |  Presentations  |  Participants  |  Funding  |  Contact
Copyright 2006 Transit Access Points (TAPs).
All Rights Reserved. Problems with this page? Contact us.
Website designed by
Transit Access Points (TAPs)
ECE Department, MS 380
Rice University
6100 Main Street
Houston, Texas 77005-1892