Limited-Feedback Resource Allocation in Heterogeneous Cellular Networks

作者：Mokari Nader; Alavi Faezeh; Parsaeefard Saeedeh^*; Tho Le Ngoc

来源：IEEE Transactions on Vehicular Technology, 2016, 65(4): 2509-2521.

DOI：10.1109/TVT.2015.2428997

• Iterative successive convex approximation (SCA)
• limited-feedback resource allocation
• Lloyd algorithm

In this paper, we develop and analyze three limited-feedback resource allocation algorithms suitable for uplink transmission in heterogeneous wireless networks (HetNets). In this setup, one macro-cell shares the spectrum with a set of underlay cognitive small-cells via the orthogonal frequency-division multiple access (OFDMA), where the interference from small-cells to the macro-cell should be kept below a predefined threshold. The resource allocation algorithms aim to maximize the weighted sum of instantaneous rates of all users over all cells by jointly optimizing power and subcarrier allocation under power constraints. Since in practice, the HetNet backhaul capacity is limited, reducing the amount of channel state information (CSI) feedback signaling passed over the backhaul links is highly desirable. To reach this goal, we apply the Lloyd algorithm to develop the limited-feedback two-phase resource allocation scheme. In the first offline phase, an optimal codebook for power and subcarrier allocation is designed and sent to all nodes. In the second online phase, based on channel realizations, the appropriate codeword of the designed codebook is chosen for transmission parameters, and the macro-cell only sends the codeword index represented by a limited number of bits for subcarrier and power allocation to its own users and small-cells. Then, each small-cell informs its own users by their related codewords. The offline phase involves a mixed-integer nonconvex resource allocation problem encountering high computational complexity. To solve it efficiently, we apply the general iterative successive convex approximation (SCA) approach, where the nonconvex optimization problem is transformed into the approximated convex optimization problem in each iteration. The simulation results reveal that the Lloyd algorithm can offer a performance close to the perfect-CSI case (without a limited number of feedback bits).

• 出版日期2016-4
• 单位McGill