Proposal Number: 0635172 Panel Summary: Panel Summary This proposal is about the study of the information rate needed to specify a time-varying MIMO channel using detailed models for the physical environment, and the associated gains in the system performance. The panel found PI's approach to the problem interesting and atypical but felt that the proposal is lacking sufficient rigourous science for the justification of the proposed action. The panel would also like to see more justification in the neccessity of the detailed modeling of the scatterers in the system over the conventional statistical model. The panel recognizes that the WINLAB programs generate very good broader impact of the conducted research. Overall, the panel rated the proposal as non-competitive. Panel Recommendation: Non-Competitive __________________________________________ Review #1 Proposal Number: 0635172 Performing Organization: Rutgers Univ New Brunswick NSF Program: Theoretical Foundations (TF) Principal Investigator: Rose, Christopher Proposal Title: How Well Can We Know A MIMO Channel? Rating: Good REVIEW: What is the intellectual merit of the proposed activity? The basic idea of this proposal is that the moving scattering objects (which is the cause of time-varying fading of MIMO channels) are all subject to a certain energy constraints, which could serve as an upper-bound for the entropy rate of CSI. This proposal hopes to establish possibly analytical relationship between the energy bound and MIMO channel CSI. The main research topics include the followings: (1) To quantify the necessary CSI for accuracy level. (2) To understand channel variation as a function of energy bounds on physical mobility of the constituent scatering objects. (3) To determine when and how the model can be obtained blindly. (4) To analyze the feasibility of channel prediction. What are the broader impacts of the proposed activity? At WINLAB the PI is focusing on training of undergraduate and graduate students for wireless communication related research. The proposed project will be used to further this mission. The proposed research may also have significant impact on understanding the MIMO channel, and thus on the design of future wireless networks. Summary Statement The proposal is well written and organized, but in a quite insightful style. The idea of finding the entropy rate bound for MIMO channel is compelling, but its realisticness needs to be justified. __________________________________________ Review #2 Proposal Number: 0635172 Performing Organization: Rutgers Univ New Brunswick NSF Program: Theoretical Foundations (TF) Principal Investigator: Rose, Christopher Proposal Title: How Well Can We Know A MIMO Channel? Rating: Good REVIEW: What is the intellectual merit of the proposed activity? The PI proposes to study the information rate needed to specify a time-varying MIMO channel based on physical models, and determine associated benefits in system performance. What are the broader impacts of the proposed activity? The proposed effort could leverage off of existing industrial ties through Winlab. Summary Statement A primary objective of the proposed work is to determine the information rate needed to describe a time-varying MIMO channel. I found the discussion interesting, and the PI has a strong track record, but I am skeptical about the benefits of the model-based approach suggested. While some means of determining whether or not it is worthwhile to bring channel information back to the transmitter can clearly be beneficial, the proposal is not very clear about how that might be accomplished in an efficient way. There has been a substantial amount of work on wireless channel prediction and estimation, which should be tied in to the proposed effort. Can low-order statistical models (e.g., ARMA) capture the time-variations associated with the proposed scattering model? If so, then it may be easier to work directly with a statistical model that implicitly and automatically accounts for the presence of scatterers. In contrast, it is not clear how the number of scatterers positions, etc, can be estimated easily. _____________________________________ Review #3 Proposal Number: 0635172 Performing Organization: Rutgers Univ New Brunswick NSF Program: Theoretical Foundations (TF) Principal Investigator: Rose, Christopher Proposal Title: How Well Can We Know A MIMO Channel? Rating: Very Good REVIEW: What is the intellectual merit of the proposed activity? The proposed activity has deep intellectual roots that penetrate the soil where the layer of engineering permeates the layer of science and the layer of philosophy. This proposal is about understanding the real world by pursuing a novel insight into the MIMO problem. What are the broader impacts of the proposed activity? The work could lead to better models for MIMO channels, and better security for MIMO networks. In this case, the links from investigation to practical application of the insights and results derived from the study are not rigid. The work could fall flat and it could end up in Nature. The thinking is very clear and perceptive. This proosal has exciting ideas. Summary Statement This proposal is atypical in that it does not give well delineated and detailed topics for investigation. Rather, it raises some fundamental questions and demonstrates some profound thinking. The discussion presented in the proposal succeeds in convincing the reader that if answers to these very basic questions can be found, the applicant is the individual to find them. I would like to see this proposal funded as a high-risk, high-reward venture. ______________________________________ Review #4 Proposal Number: 0635172 Performing Organization: Rutgers Univ New Brunswick NSF Program: Theoretical Foundations (TF) Principal Investigator: Rose, Christopher Proposal Title: How Well Can We Know A MIMO Channel? Rating: Good REVIEW: What is the intellectual merit of the proposed activity? The proposed research aims to answer the fundamental question: Is it possible to learn the coefficients of a given MIMO fading channel through interrogation? The approach is to derive bounds on the amount of information necessary to describe the motion of relevant channel constituents. What are the broader impacts of the proposed activity? Learning and tracking fading channels are crucial for harnessing the diversity and multiplexing gains promised by MIMO. The questions posed in the proposal, if answered successfully, will have great impact in the design of practical MIMO systems. The impact also includes training graduate students and technology transfer. Summary Statement The PI describes radical ideas not often seen in the research community. The question of how well we can learn a MIMO channel and how much information is necessary to describe such a channel are of fundamental importance. The approach of comparing the amount of information necessary for describing channel constituents and the amount of information that can be extracted from channel interrogation is very interesting. The proposal is to a large extent mental excercises with few physical evidence. Some of the ideas are subject to doubt. For example, the number of scatterers can be extremely large, even if each scatterer is simple (a single mass such as a tree can contribute to a large number of paths). It is in feasible to describe the motion of foliage or similarly complicated objects. The proposal can be strengthened if the PI plans also experimental study to verify the results, or enlist an experimentalist to help.