Backhaul load and performance optimality of partial joint processing schemes in LTE-A networks

Surging demand for more new mobile broadband services everywhere, urged 3GPP to take advantage of small cells and frequency reuse factor of one in 4G LTE. However, the problem of performance degradation caused by InterCell-Interference (ICI) also raised for cell-edge users. Coordinated Multipoint Transmission and Reception (COMP) is a cutting edge solution to ICI in LTEAdvanced. COMP techniques are divided into Coordinated Scheduling / Beamforming and Joint Processing. This thesis focuses on downlink joint processing, where each user receives data from various transmission points, improving the signal strength and cancelling interference. Joint processing transmission strategy is further divided into two schemes: Joint Transmission (JT) in which, multiple points transmit simultaneously to a user and Dynamic Point Selection (DPS) where, a single transmitter accommodates a user at each Transmission Time Interval (TTI). Joint processing demands for considerable amount of backhauling, signaling and data sharing that put a doubt on its feasibility. Previously, Partial Joint Processing has been proposed and evaluated for joint transmission as the most demanding scheme for backhaul in a static cluster for a flat fading Rayleigh channel. By limiting the cooperation in an active set of Evolved Node B (eNBs) for each user, the feedback and backhaul load is reduced with some performance degradation. In this thesis, Centralized and Partial Joint Processing are mathematically defined and evaluated for both JT and DPS in a multi-path environment using the WINNER II channel model which has been developed for LTE-Advanced. Centralized means full cooperation when complete channel matrix presents in the beamformer. But, partial joint processing translates in sparse channel matrices available at the central precoder entity and leads in the rank deficiency of channel correlation matrix in the zero forcing beamformer (ZFBF). A dynamic user-wise algorithm is proposed to resolve this problem. In depth comparison among these schemes using different metrics like average sum-rate per cell, data rate and feedback rate demonstrates the better performance of centralized cooperation over partial cooperation with higher backhaul load. Joint transmission outperforms dynamic point selection but with higher backhaul requirements. The utility of PJP schemes for three different traffic types shows that DPS is preferred over JT for elastic and adaptive applications. The significance of the proposed algorithm is also proven in comparison with previous approach by maintaining the same performance while reducing the backhaul load up to 30%. The proposed algorithm also leverages the location dependency of the joint processing. All these achieved because the proposed algorithm targets those users located at the cluster borders who more likely cause rank deficiency. Keywords: Inter-Cell Interference, 3GPP, CoMP, 4G LTE, Partial Joint Processing, WINNER, Precoder, Beamforming, Zero-Forcing.

Preview Text FK 2014 130 edited IR.pdf Download (1MB) | Preview

Actions (login required)

View Item