Final Report Summary - CODING FOR NETWORKS (Coding for Multi-User Channels for Next Generation Wireless Networks)
Wireless networking constitutes an important component of future information technology applications ranging from ad-hoc and sensor networks to cellular networks with a very strong infrastructure, or even body area networks. In existing designs, interference is either avoided by a carefully chosen networking protocol (e.g. through time-division, space-division, frequency-division), or it is controlled to be near an acceptable level (e.g. through code-division). Despite considerably simplifying the design and implementation of wireless communication networks, the current approaches (which boil down to nothing but point-to-point links at the physical layer) are highly sub-optimal as they do not exploit “multi-user communication links” at the physical layer, necessary for optimal performance. As a result, for reliable and efficient wireless networks of the future, a different look at the problem of handling the interference is needed.
Existing research in multi-user communications from a physical-layer perspective is dominated by information theoretic results. Especially, since late 1990’s many important information theoretic advancements to determine the limits of communication over wireless channels have been obtained. On the other hand, complementary results on practical signaling solutions are largely missing. With this motivation, through this research project, we aim to develop practical channel coding and modulation solutions enabling implementation of efficient wireless networks that exploit interference in an optimal manner. Specifically, we are interested in signaling for interference channels and two-way relay channels which can be considered as some of the building blocks for a strong physical layer for future wireless communication systems. In addition to contributing to future wireless networking products, the project involves educational elements (such as graduate student education) and public outreach activities (such as exposing high school students to wireless communications related research and development activities).
Towards the above goals, during the course of this four-year project, state of the art code designs (e.g. through the use and design of low density parity check codes) for Gaussian interference channels have been completed. Also, robust signaling solutions for the two-way relay channels exhibiting time and frequency selective fading are designed. Further, short block length code designs for some simple multi-user set-ups are considered and novel coding/modulation solutions are obtained. These results are published in eleven journal papers in the top outlets of the communications field (almost all in IEEE Transactions), and seven conference papers (one of which is invited) in top conferences.
The main impact of the work is scientific and technical. Comparing to the previously existing results in the literature:
1- We now have a method of practical channel coding/modulation for use implementing codes directly for the interference channel. Earlier systems/solutions employed time or frequency division type techniques, basically operating far from the theoretically best limits. We now can operate close to the information theoretic limits demonstrating that the designed low density parity check codes are near optimal, and should be employed for a more efficient use of the power and bandwidth resources (compared to the existing solutions). Methods are developed for both static and fading channels.
2- We now have an efficient solution for solving the synchronization problem for two-way relay systems (that can be used in wireless radio communications, or in underwater acoustic channels). Our approach is superior to the existing solutions since it does not use long cyclic prefixes, instead it creates a delay diversity system, and extracts the best performance achievable. Our solution outperforms the existing ones in achievable data rates and in terms of the error probabilities.
3- We now have a method of designing short length codes (for delay sensitive applications) for use over multiple access channels. Our findings will form the fundamentals of the work to be done in additional multi-user settings, and clearly demonstrate that using trellis based codes is a better than other "capacity achieving" codes when the block lengths are short.
4- We now have a method of analysing the performance of insertion/deletion channels, and reported the results in two journal papers: one in IEEE Trans. on Information Theory and the other in IEEE Trans. on Communications. These findings represent fundamental research accomplishments describing previously unknown results, e.g. demonstrating a tight upper bound (better than the existing results) for the deletion channel capacity. This is a very hard problem being studied for over 50 years, and the PI has obtained the best available bounds on certain cases in the entire literature - which is a significant scientific impact.
To summarize: The project directly contributed toward the re-integration of the Marie Curie Fellow Tolga M. Duman to Europe. After spending nearly twenty years, first as a graduate student and then as a faculty member in the United States, Prof. Dr. Duman has returned to Bilkent University in Turkey as a member of the Department of Electrical and Electronics Engineering in September 2012. In part, through the support of this project, he was able to establish an externally funded research group at Bilkent University working towards their M.S. and Ph.D. degrees. His group's website can be accessed at http://ctar.bilkent.edu.tr. His research group is currently composed of four Ph.D. students and four M.S. students, and several admission offers are made for the fall 2017 term to recruit new students. He has already graduated three M.S. students (with thesis), and three other M.S. students are scheduled to graduate during the 2017 summer term. Two of the Ph.D. students have just completed their fourth years, and they are expected to graduate within the next academic year.
Existing research in multi-user communications from a physical-layer perspective is dominated by information theoretic results. Especially, since late 1990’s many important information theoretic advancements to determine the limits of communication over wireless channels have been obtained. On the other hand, complementary results on practical signaling solutions are largely missing. With this motivation, through this research project, we aim to develop practical channel coding and modulation solutions enabling implementation of efficient wireless networks that exploit interference in an optimal manner. Specifically, we are interested in signaling for interference channels and two-way relay channels which can be considered as some of the building blocks for a strong physical layer for future wireless communication systems. In addition to contributing to future wireless networking products, the project involves educational elements (such as graduate student education) and public outreach activities (such as exposing high school students to wireless communications related research and development activities).
Towards the above goals, during the course of this four-year project, state of the art code designs (e.g. through the use and design of low density parity check codes) for Gaussian interference channels have been completed. Also, robust signaling solutions for the two-way relay channels exhibiting time and frequency selective fading are designed. Further, short block length code designs for some simple multi-user set-ups are considered and novel coding/modulation solutions are obtained. These results are published in eleven journal papers in the top outlets of the communications field (almost all in IEEE Transactions), and seven conference papers (one of which is invited) in top conferences.
The main impact of the work is scientific and technical. Comparing to the previously existing results in the literature:
1- We now have a method of practical channel coding/modulation for use implementing codes directly for the interference channel. Earlier systems/solutions employed time or frequency division type techniques, basically operating far from the theoretically best limits. We now can operate close to the information theoretic limits demonstrating that the designed low density parity check codes are near optimal, and should be employed for a more efficient use of the power and bandwidth resources (compared to the existing solutions). Methods are developed for both static and fading channels.
2- We now have an efficient solution for solving the synchronization problem for two-way relay systems (that can be used in wireless radio communications, or in underwater acoustic channels). Our approach is superior to the existing solutions since it does not use long cyclic prefixes, instead it creates a delay diversity system, and extracts the best performance achievable. Our solution outperforms the existing ones in achievable data rates and in terms of the error probabilities.
3- We now have a method of designing short length codes (for delay sensitive applications) for use over multiple access channels. Our findings will form the fundamentals of the work to be done in additional multi-user settings, and clearly demonstrate that using trellis based codes is a better than other "capacity achieving" codes when the block lengths are short.
4- We now have a method of analysing the performance of insertion/deletion channels, and reported the results in two journal papers: one in IEEE Trans. on Information Theory and the other in IEEE Trans. on Communications. These findings represent fundamental research accomplishments describing previously unknown results, e.g. demonstrating a tight upper bound (better than the existing results) for the deletion channel capacity. This is a very hard problem being studied for over 50 years, and the PI has obtained the best available bounds on certain cases in the entire literature - which is a significant scientific impact.
To summarize: The project directly contributed toward the re-integration of the Marie Curie Fellow Tolga M. Duman to Europe. After spending nearly twenty years, first as a graduate student and then as a faculty member in the United States, Prof. Dr. Duman has returned to Bilkent University in Turkey as a member of the Department of Electrical and Electronics Engineering in September 2012. In part, through the support of this project, he was able to establish an externally funded research group at Bilkent University working towards their M.S. and Ph.D. degrees. His group's website can be accessed at http://ctar.bilkent.edu.tr. His research group is currently composed of four Ph.D. students and four M.S. students, and several admission offers are made for the fall 2017 term to recruit new students. He has already graduated three M.S. students (with thesis), and three other M.S. students are scheduled to graduate during the 2017 summer term. Two of the Ph.D. students have just completed their fourth years, and they are expected to graduate within the next academic year.