Distributed Space–Time Interference Alignment With Moderately Delayed CSIT

Abstract

Abstract-This paper proposes an interference alignment method with distributed and delayed channel state information at the transmitter (CSIT) for a class of interference networks. The core idea of the proposed method is to align interference signals over time at the unintended receivers in a distributed manner. With the proposed method, achievable trade-offs between the sum of degrees of freedom (sum-DoF) and feedback delay of CSI are characterized in both the X-channel and three-user interference channel to reveal the impact on how the CSI feedback delay affects the sum-DoF of the interference networks. A major implication of derived results is that distributed and moderatelydelayed CSIT is useful to strictly improve the sum-DoF over the case of no CSI at the transmitter in a certain class of interference networks. For a class of X-channels, the results show how to optimally use distributed and moderately-delayed CSIT to yield the same sum-DoF as instantaneous and global CSIT. Further, leveraging the proposed transmission method and the known outer bound results, the sum-capacity of the two-user X-channel with a particular set of channel coefficients is characterized within a constant number of bits.

Degrees of freedom of three-user SISO interference channels under hybrid CSI at the transmitters

“…We show that a DoF of 1.2 can be attained with our hybrid CSIT scheme which is strictly greater than the DoF of 36 31 achieved with completely delayed CSIT in [14]. Please note that very recently, a hybrid CSIT based scheme has been proposed for the 3-user SISO-IC in [18] which also achieves a DoF of 1.2. While the second phase of our scheme is inspired from [18], our result is different from [18] in 2 important ways.…”

“…While the second phase of our scheme is inspired from [18], our result is different from [18] in 2 important ways. Firstly, in [18], a DoF of 1.2 is achieved where local instantaneous CSIT (LI-CSIT) should be available at least 40% of the channel coherence time whereas in our proposed scheme, LI-CSIT should be available only 20% of the channel coherence time. This implies that our scheme can tolerate more feedback delays (such as in faster time-varying channels) and yet obtain a DoF gain of 1.2.…”

“…Please note that very recently, a hybrid CSIT based scheme has been proposed for the 3-user SISO-IC in [18] which also achieves a DoF of 1.2. While the second phase of our scheme is inspired from [18], our result is different from [18] in 2 important ways. Firstly, in [18], a DoF of 1.2 is achieved where local instantaneous CSIT (LI-CSIT) should be available at least 40% of the channel coherence time whereas in our proposed scheme, LI-CSIT should be available only 20% of the channel coherence time.…”

“…This implies that our scheme can tolerate more feedback delays (such as in faster time-varying channels) and yet obtain a DoF gain of 1.2. Secondly, in [18], some of the messages are "common" in nature, i.e., each user can decode some of the messages of other users also. However, in our scheme, using the concept of second-order message generation from [14], each user's message is strictly private.…”

Achieving the DoF limits of the SISO X channel with imperfect-quality CSIT

“…Motivated by works such as that in [11] which presented a distributed interference management technique which can obtain the optimal DoF with local and perfect current CSIT with a certain fractional delay, and by works on imperfect current and delayed CSIT over the broadcast channel [6]- [9], we here explore the role of feedback in moving between these extremal points (α = 1 and β = 1, α = 0) by considering different values of α and β.…”