Multiple Cell Limited Feedback
Co-channel interference (CCI), arising due to simultaneous transmissions by neighboring base stations can cause outages and reduce data rates in cellular systems, particularly at the cell-edges. CCI can be managed by multicell cooperation, which involves base stations coordinating their transmissions by exchanging user information over a backhaul link. More information on multicell base station cooperation can be found here.
User channel state information (CSI) can be exchanged among base stations to obtain the gains promised by cooperative strategies. CSI can be made available at the base stations using finite-bandwidth feedback channels. Limited feedback for single- and multi-user MIMO systems is described here and here, respectively. Existing literature on non-cooperative limited feedback is, however, not directly applicable to multicell cooperative systems. In single-cell systems, the CSI of only the desired channel is necessary at the transmitter side, while base stations in cooperative networks require the CSI of multiple channels (desired + interfering). This implies that each user has to feedback the CSI of multiple channels over the finite-bandwidth channel. Hence, the available feedback resources need to be partitioned among different channels. This is illustrated in the figure shown.
Since the uplink bandwidth is limited, it is important that it is utilized efficiently. Thus, splitting resources equally among the multiple channels is not necessarily the best solution. For example, consider the case when there are B feedback bits available and K channels to quantize. Then, if the interfering channels are too weak to affect some performance metric (like data-rate), it makes most sense to allocate all B bits to quantize and feedback only the desired channel with higher resolution, instead of expending (K-1)B/K bits on quantizing the (K-1) weak interfering channels. A similar argument can be made in terms of the delay in feedback and backhaul links – channels with large delays can be assigned fewer bits than those with smaller delays. Hence, the available bandwidth is most efficiently utilized by partitioning adaptively the feedback resources among different channels, as a function of the signal strengths and delays.
In WSIL, we have been working on proposing novel multicell limited feedback approaches. We have developed different techniques like separate quantization (where the available B bits are partitioned adaptively among multiple channels), joint quantization (where all B bits are used to quantize a joint channel vector, by introducing adaptive biases to feedback stronger channels with more resolution), and predictive joint quantization (where temporal correlation among channels is exploited to reduce feedback requirements).
R. Bhagavatula and R. W. Heath, Jr., “Adaptive Limited Feedback for Sum-Rate Maximizing Beamforming in Cooperative Multicell Systems” accepted to the IEEE Trans. on Signal Processing, October 2010.
In this paper, a single-interferer multicell system is considered. A new limited feedback strategy is developed to partition the available bits between the desired and interfering channels, to reduce the mean loss in sum-rate due to quantization. The proposed feedback-bit partitioning algorithm is shown, using simulations, to yield significant gains in the sum-rates using limited feedback.
R. Bhagavatula and R. W. Heath, Jr., “Adaptive Bit Partitioning for Multicell Intercell Interference Nulling with Delayed Limited Feedback,” submitted to the IEEE Trans. on Signal Processing, July 2010.
This paper considers multicell systems where each user faces interference from multiple interferers and the feedback and backhaul links are associated with delays. Closed-form expressions are derived for feedback-bit partitions to allocate bits as a function of the signal strengths and delays of the channels. Simulations are used to conclude that the proposed algorithm yields much higher sum-rates as compared to equal-bit assignment.
R. Bhagavatula, R. W. Heath, Jr., and B. Rao “Limited Feedback with Joint CSI Quantization for Multicell Cooperative Generalized Eigenvector Beamforming,” Proc. of the IEEE Int. Conf. on Acoustics, Speech, and Signal Proc., Dallas, TX, pp. 2838-2841, March 14-19, 2010.
It is proposed to feedback all the channels jointly as a net channel vector, and a selection criterion is derived to quantize stronger channels with finer resolution. Jointly quantizing all the channels in a multicell system yields higher data rates as compared to separate quantization approaches.
The research in the area of multicell limited feedback was funded through NSF grants, particularly, NSF-CCF-0514194 and NSF-CCF-0830615.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.