Final Report Summary - RFC (Cooperation Strategies in Wireless Networks: Relaying, Feedback and Conferencing)
Motivation and objectives:
--------------------------------------
In recent years there has been a sharp increase in the demand for reliable, high-rate wireless communications. Cooperative communications is a central approach for improving the performance of wireless networks aiming to cope with the increasing demand. The underlying idea in cooperative communications is to manipulate the interference in novel ways to simultaneously improve the performance of the nodes in the network. It thus follows that cooperative communications constitutes an alternative paradigm to the currently ruling paradigm, which considers interference as a harmful phenomenon that should be avoided. Broadly speaking, cooperation strategies can be classified into three types: relaying, feedback and conferencing. The objective of this project is to investigate the application of cooperative strategies in the presence of interference.
Work done to date and main results (overall project duration)
--------------------------------------------------------------------------------------
Project # 1: The Value of Cooperation Between Relays in the Multiple Access Channel With Multiple Relays.
In this project we studied the discrete, memoryless multiple-access channel with two independent sources, two relays, and a single destination. We refer to this configuration as the multiple-access channel with multiple relays (MACMR), which is a generalization of the multiple-access relay channel (MARC) model obtained by adding a relay node. We presented inner and outer bounds on the capacity region of the MACMR. The inner bound is based on a hierarchical decode-and-forward scheme, in which each relay decodes the messages of the lower hierarchy. We extend the regular encoding, sliding-window decoding and backward decoding techniques, previously applied to MARCs and multiple-relay channels, to MACMRs. The outer bounds are obtained using the cut-set-bound. For Gaussian MACMRs the bounds were evaluated and compared to those obtained for the multiple-access channel with parallel relays. We demonstrated that significant improvement in the performance can be obtained by letting relays interact with each other.
Project # 2: The Capacity Region of the Fading Interference Channel with a Relay in the Strong Interference Regime.
The interference channel with a relay (ICR) consists of two pairs communicating over a shared wireless channel, and a relay, whose objective is to simultaneously assist both communicating pairs. The ICR is the fundamental building block of cooperation in wireless networks which includes multiple communicating pairs interfering with each other. In this project, ICRs in which the links are subject to i.i.d. fading, were considered. Specifically, two channel models were studied, phase-fading and Rayleigh fading. Strong interference conditions were derived for the case where the links from the sources to the relay are good, leading to the characterization of the capacity region for such scenarios. This is the first time the capacity region of this model is characterized for a non-degraded, non-cognitive scenario, with a causal relay when all links are active. Several important conclusions resulted from this work: First, it was shown that a single relay can optimally assist several communicating pairs simultaneously. This implies that the number of relays needed to improve performance is wireless networks is less than the number of pairs. Second, the optimal codebooks were shown to be mutually independent. This means that the sources do not need to be aware of the presence of the relay – only the destination needs to know. This greatly simplifies network management. Furthermore, as the sources do not need to change their codebooks, it simplifies the integration of relaying into existing systems. Therefore, our conclusions strongly support the application of relaying for interference management in fading wireless networks.
Project # 3: Capacity of the Fading Interference Channel with a Relay and Feedback Links.
In this project we studied the simultaneous application of both relaying and feedback, which are the most important cooperation types, in the presence of interference. We obtained exact characterization of the capacity regions for Rayleigh fading and phase fading interference channels with a relay and with feedback links, in the strong and very strong interference regimes. Four feedback configurations are considered: (1) feedback from both receivers to the relay, (2) feedback from each receiver to the relay and to one of the transmitters (either corresponding or opposite), (3) feedback from one of the receivers to the relay, (4) feedback from one of the receivers to the relay and to one of the transmitters. The benefits of each feedback configurations were characterized. It was shown that a single relay can optimally assist both communicating pairs simultaneously, also in the presence of feedback, and that the optimal codebooks are mutually independent. Therefore, our results show that there is a strong motivation for incorporating relaying and feedback for interference management in wireless networks.
Project # 4: Source-Channel Coding Theorems for the Multiple-Access Relay Channel.
In the fourth project we studied reliable transmission of arbitrarily correlated sources over multiple-access relay channels (MARCs) and multiple-access broadcast relay channels (MABRCs). In MARCs only the destination is interested in a reconstruction of the sources, while in MABRCs both the relay and the destination want to reconstruct the sources. In addition to arbitrary correlation among the source signals at the users, both the relay and the destination are assumed to have side information correlated with the source signals. Sufficient conditions for reliable communication based on operational separation, as well as upper bounds on the achievable source-channel rate are characterized. Since the operational separation for MARCs and MABRCs is not optimal in general, sufficient conditions for reliable communication using a joint source-channel coding scheme based on the correlation preserving mapping technique are also found. For correlated sources transmitted over fading Gaussian MARCs and MABRCs, conditions under which informational separation is optimal are found.
Project # 5: The generalized degrees-of-freedom of the ICR.
As mentioned earlier, the importance of the ICR follows since it is the fundamental building block of cooperation in wireless networks. The research of the ICR started in the first period of the project (see projects 2 and 3), where we studied the ergodic ICR with and without feedback, in the strong and the very strong interference regimes. In this project the focus was on the ICR at the asymptotically high SNR regime. We first obtained the generalized degrees-of-freedom (GDOF) of the Z-ICR (i.e. the ICR in which only one receiver suffers from interference), in which the relay overhears transmissions of only one sender. We showed that when interference is weak, then the decode-and-forward relaying scheme is optimal in the GDOF sense. We also showed that the relay supports a higher GDOF compared to the Z-IC (i.e. the Z interference channel without a relay). Combined with the results of projects 2 and 3 these results provide very strong case for using cooperation for interference management in wireless networks.
Project # 6: Joint source-channel coding (JSCC) over broadcast channels with causal feedback.
In project # 4 (in the reporting period) we studied JSCC for multiple-access relay channels, where we studied the effect of relay-type cooperation on the achievable source-channel rates. In this project we study the effect of feedback-type cooperation on the achievable source-channel rates in the broadcast channel. Specifically, we focus on uncoded transmission of correlated sources in the finite horizon regime. We analyzed the finite-horizon performance of the Ozarow-Leung (OL) scheme and of the LQG scheme, and showed that while the LQG scheme is superior in the infinite horizon regime, in the finite-horizon regime there are scenarios in which the OL scheme allows a better delay-distortion tradeoff.
Project # 7: Practical coding schemes for multiple-access relay channels.
In this project we constructed a novel practical full-duplex cooperative coding scheme for the MARC (by “practical coding scheme” we mean that the scheme is based on practically implementable codes and the transmitted symbols are taken from finite constellations). The main novelty of this work is the introduction of an algorithm for generating the information to be transmitted from the relay, whose objective is to assist in the joint decoding of the codewords of both senders at the destination. In this project we completed the code construction and initial performance analysis. We are currently working on the detailed performance characterization.
--------------------------------------
In recent years there has been a sharp increase in the demand for reliable, high-rate wireless communications. Cooperative communications is a central approach for improving the performance of wireless networks aiming to cope with the increasing demand. The underlying idea in cooperative communications is to manipulate the interference in novel ways to simultaneously improve the performance of the nodes in the network. It thus follows that cooperative communications constitutes an alternative paradigm to the currently ruling paradigm, which considers interference as a harmful phenomenon that should be avoided. Broadly speaking, cooperation strategies can be classified into three types: relaying, feedback and conferencing. The objective of this project is to investigate the application of cooperative strategies in the presence of interference.
Work done to date and main results (overall project duration)
--------------------------------------------------------------------------------------
Project # 1: The Value of Cooperation Between Relays in the Multiple Access Channel With Multiple Relays.
In this project we studied the discrete, memoryless multiple-access channel with two independent sources, two relays, and a single destination. We refer to this configuration as the multiple-access channel with multiple relays (MACMR), which is a generalization of the multiple-access relay channel (MARC) model obtained by adding a relay node. We presented inner and outer bounds on the capacity region of the MACMR. The inner bound is based on a hierarchical decode-and-forward scheme, in which each relay decodes the messages of the lower hierarchy. We extend the regular encoding, sliding-window decoding and backward decoding techniques, previously applied to MARCs and multiple-relay channels, to MACMRs. The outer bounds are obtained using the cut-set-bound. For Gaussian MACMRs the bounds were evaluated and compared to those obtained for the multiple-access channel with parallel relays. We demonstrated that significant improvement in the performance can be obtained by letting relays interact with each other.
Project # 2: The Capacity Region of the Fading Interference Channel with a Relay in the Strong Interference Regime.
The interference channel with a relay (ICR) consists of two pairs communicating over a shared wireless channel, and a relay, whose objective is to simultaneously assist both communicating pairs. The ICR is the fundamental building block of cooperation in wireless networks which includes multiple communicating pairs interfering with each other. In this project, ICRs in which the links are subject to i.i.d. fading, were considered. Specifically, two channel models were studied, phase-fading and Rayleigh fading. Strong interference conditions were derived for the case where the links from the sources to the relay are good, leading to the characterization of the capacity region for such scenarios. This is the first time the capacity region of this model is characterized for a non-degraded, non-cognitive scenario, with a causal relay when all links are active. Several important conclusions resulted from this work: First, it was shown that a single relay can optimally assist several communicating pairs simultaneously. This implies that the number of relays needed to improve performance is wireless networks is less than the number of pairs. Second, the optimal codebooks were shown to be mutually independent. This means that the sources do not need to be aware of the presence of the relay – only the destination needs to know. This greatly simplifies network management. Furthermore, as the sources do not need to change their codebooks, it simplifies the integration of relaying into existing systems. Therefore, our conclusions strongly support the application of relaying for interference management in fading wireless networks.
Project # 3: Capacity of the Fading Interference Channel with a Relay and Feedback Links.
In this project we studied the simultaneous application of both relaying and feedback, which are the most important cooperation types, in the presence of interference. We obtained exact characterization of the capacity regions for Rayleigh fading and phase fading interference channels with a relay and with feedback links, in the strong and very strong interference regimes. Four feedback configurations are considered: (1) feedback from both receivers to the relay, (2) feedback from each receiver to the relay and to one of the transmitters (either corresponding or opposite), (3) feedback from one of the receivers to the relay, (4) feedback from one of the receivers to the relay and to one of the transmitters. The benefits of each feedback configurations were characterized. It was shown that a single relay can optimally assist both communicating pairs simultaneously, also in the presence of feedback, and that the optimal codebooks are mutually independent. Therefore, our results show that there is a strong motivation for incorporating relaying and feedback for interference management in wireless networks.
Project # 4: Source-Channel Coding Theorems for the Multiple-Access Relay Channel.
In the fourth project we studied reliable transmission of arbitrarily correlated sources over multiple-access relay channels (MARCs) and multiple-access broadcast relay channels (MABRCs). In MARCs only the destination is interested in a reconstruction of the sources, while in MABRCs both the relay and the destination want to reconstruct the sources. In addition to arbitrary correlation among the source signals at the users, both the relay and the destination are assumed to have side information correlated with the source signals. Sufficient conditions for reliable communication based on operational separation, as well as upper bounds on the achievable source-channel rate are characterized. Since the operational separation for MARCs and MABRCs is not optimal in general, sufficient conditions for reliable communication using a joint source-channel coding scheme based on the correlation preserving mapping technique are also found. For correlated sources transmitted over fading Gaussian MARCs and MABRCs, conditions under which informational separation is optimal are found.
Project # 5: The generalized degrees-of-freedom of the ICR.
As mentioned earlier, the importance of the ICR follows since it is the fundamental building block of cooperation in wireless networks. The research of the ICR started in the first period of the project (see projects 2 and 3), where we studied the ergodic ICR with and without feedback, in the strong and the very strong interference regimes. In this project the focus was on the ICR at the asymptotically high SNR regime. We first obtained the generalized degrees-of-freedom (GDOF) of the Z-ICR (i.e. the ICR in which only one receiver suffers from interference), in which the relay overhears transmissions of only one sender. We showed that when interference is weak, then the decode-and-forward relaying scheme is optimal in the GDOF sense. We also showed that the relay supports a higher GDOF compared to the Z-IC (i.e. the Z interference channel without a relay). Combined with the results of projects 2 and 3 these results provide very strong case for using cooperation for interference management in wireless networks.
Project # 6: Joint source-channel coding (JSCC) over broadcast channels with causal feedback.
In project # 4 (in the reporting period) we studied JSCC for multiple-access relay channels, where we studied the effect of relay-type cooperation on the achievable source-channel rates. In this project we study the effect of feedback-type cooperation on the achievable source-channel rates in the broadcast channel. Specifically, we focus on uncoded transmission of correlated sources in the finite horizon regime. We analyzed the finite-horizon performance of the Ozarow-Leung (OL) scheme and of the LQG scheme, and showed that while the LQG scheme is superior in the infinite horizon regime, in the finite-horizon regime there are scenarios in which the OL scheme allows a better delay-distortion tradeoff.
Project # 7: Practical coding schemes for multiple-access relay channels.
In this project we constructed a novel practical full-duplex cooperative coding scheme for the MARC (by “practical coding scheme” we mean that the scheme is based on practically implementable codes and the transmitted symbols are taken from finite constellations). The main novelty of this work is the introduction of an algorithm for generating the information to be transmitted from the relay, whose objective is to assist in the joint decoding of the codewords of both senders at the destination. In this project we completed the code construction and initial performance analysis. We are currently working on the detailed performance characterization.