Hydra – A Wireless Multihop Testbed
Wireless ad hoc networks provide rapidly deployable, self-configuring networks which operate without infrastructure. These networks will have far reaching applications in various areas, such as military applications (e.g. battlefield scenarios) and commercial applications (e.g. connectivity-everywhere, extending cellular network range). Multiple antennas can be used in these networks to not only increase capacity, but also to improve reliability and spatial reuse.
In order to investigate multiple-input-multiple-output (MIMO) ad hoc networking, we have developed Hydra, a flexible wireless networking testbed. The development of this wireless testbed is a collaboration between the University of Texas (UT) at Austin and Drexel University. At UT Dr. Robert Heath and his students are primarily responsible for the multiple antenna Physical layer (PHY), and Dr. Scott Nettles and his students are responsible for the Medium-Access Control (MAC), network, and application layers. At Drexel Dr. Kapil Dandekar and his students have developed the RF front end and its associated controls. The underlying motivation in the design of this flexible prototype is that students should have a mechanism for testing theoretical work with practical implementations, thus bridging the gap between theory and practice.
Hydra Nodes in the WNCG labs
Our primary design goal was to make Hydra as flexible, reconfigurable, and easy-to-modify as possible in order to maximize the utility of the prototype for researchers. To this end the PHY and MAC have been implemented in reconfigurable software. The PHY is implemented in National Instruments LabVIEW, and the MAC layer is implemented using a reconfigurable router development tool called Click. The RF front end also reflects this goal of flexibility as it is implemented using frequency agile RF transceiver cards from Texas Instruments (RCS 110). Although the use of this variety of tools requires a significant amount of equipment, more compact off-the-shelf components cannot provide us with the degree of flexibility needed for the state-of-the-art research we wish to implement.
From the Ivory Tower to the Real World
The primary motivation for the creation of Hydra is the prevailing gap between theoretical results generated by researchers and practical systems designed by engineers. Most research generated in universities is supported by analytical work and simulation. In order to make analysis/simulation tractable, it is necessary to make assumptions which are often difficult to maintain in real systems (e.g. perfect synchronization, perfect channel estimation, etc.). Prototyping allows us to study the effect of breaking these assumptions. Hydra will allow researchers to see how the theory they have generated holds up in the real world of the prototype; thereby adding even more depth and insight to their research.
Although the Hydra project is still in its infancy, we have already made great strides. We have verified the end-to-end application layer functionality of the prototype through point-to-point testing. The average throughput of the system is currently limited by the processing time needed for some of our PHY algorithms. We are speeding up these processing bottlenecks by reimplementing some of the PHY in efficient C code. Also, we are implementing remote demo/development capability through virtual network computing (VNC).
Even with its previously mentioned limited throughput, Hydra is still useful as a research tool. The first results we are pursuing using Hydra are in the areas of cross-layer design and fundamental MIMO research. We wish to benchmark rate adaptive MAC protocols such as the Receiver-Based Auto Rate (RBAR) and Opportunistic Auto Rate (OAR) protocols. We also wish to implement research by Dr. Heath and his students on closed-loop MIMO communication.