Cognition, Collaboration and Competition in Space and Time: From Theory To Implementation
Mathematics of Relaying and Cooperation in Communication Networks
April 09,2006 11:00 AM to 11:30 AM
Speakers:
Tarokh, Vahid
|
 |
Abstract: |
Elements/clusters of a sensor network operating on the same band canoperate using three different paradigms: i) Competition: This is information theoretically casted in the framework of interference channels. ii) Collaboration: Silent transmitters/receivers can help active transmitters/receivers in the transmission/reception of their messages, but have to extract this message from the underlying transmission or by other methods, and iii) Cognitive Radio Transmission: Some devices extract the message of other transmitters from their signals or by other methods, and use it tominimize interference from/to their own transmitted signals. Competition has been well-studied in the literature. Collaboration has been less studied and Cognitive Radio Transmission has not been much studied. For the case of collaboration, we demonstrate that most of the MIMO space-time gain can also be achieved through collaborative communications with single-antenna/multiple-antenna nodes when there is one receiving agent. In particular, for the single antenna case, we consider communication to take place between clusters of nearby nodes. We show the existence of collaborative codes for communications for which the intra-cluster negotiation penalty is in principle small and almost all the diversity gain of traditional space-time codes may be realized. For example, for a single transmitter node with two collaborators and one receiver node, if the collaborators have as little as 10 dB pathloss advantage over the receiver, the penalty for collaboration over traditional space-time systems is negligible. For the cognitive radio transmission, we consider a channel defined as an n-transmitter, m-receiver interference channel in which sender i obtains (causally or uncasually) the messages senders 1 through i-1 plan to transmit. The two sender, two receiver case is considered. In this scenario, one user, a cognitive radio, obtains (genie assisted,or causally) knowledge of the data to be transmitted by the other user. The cognitive radio may then simultaneously transmit over the same channel, as opposed to waiting for an idle channel as in a traditional cognitive radio channel protocol. Dirty-paper coding and ideas from achievable region constructions for the interference channel are used, and an achievable region for the cognitive radio channel is computed. It is shown that in the Gaussian case, the described achievable region approaches the upper bounds provided by the 2x2 Gaussian MIMO broadcast channel, and an interference-free channel. Finally, we demonstrate/discuss the implementation of these ideas on an SDR platform.
This is a joint work with Natasha Devroye, Patrick Mitran, Hideki Ochiai (Harvard) and John Chapin's Team (Vanu Systems). |
|
|
