Research On Radio Resource Allocation Strategy For Macro-Femto Networks Based On Cross-Layer Design

Compared to the traditional cellular network architecture, the Macro-Femto architecture can improve the coverage and increase the system capacity. The challenges Macro-Femto network face include maximizing the system throughput, minimizing the system energy consumption, guaranteeing quality of service (QoS) of different services, etc. Recent years, the cross-layer design has been paid much attention due to its ability of improving the system performance, reducing the interference, and decreasing the power consumption. It’s timely for us to study the radio resource allocation strategy based on cross-layer design in Macro-Femto Networks with QoS provisions.The disadvantages of existing cross-layer technologies are shown as follows. Ignore the cross-layer design covering completely from the physical to the application layer, which will lead to degradation in overall system performance. The traffic model is not practical. In realistic environment, each user has multiple heterogeneous applications simultaneously. Multiple input multiple output-orthogonal frequency divison multiplexing (MIMO-OFDM) system based on cross-layer design is not studied thoroughly. There has been very few research works addressing the issues on the radio resource allocation based on cross-layer design specifically for Macro-Femto Networks. Most of the existing cross-layer designs heavily rely on perfect channel state information at the base station (CSIT), which is difficult to obtain in practice. A systematic framework is proposed to address the above design issues in Macro-Femto Networks.The cross-layer optimization problem is formulated which aims to maximize the system utility, and guarantee the QoS of different kinds of services at the same time. This model can efficiently characterize the interactions across five layers. The constraints considered include:(1) the power allocation and the radio resource block (RB) allocation in the physcial layer.(2) the scheduling scheme in the medium access control (MAC) layer.(3) the routing strategy in the network layer.(4) the reliable transmission scheme in the transport layer.(5) the QoS requirements in the application layer. The original problem is a mixed integer non-linear programming problem. Exhaustive search is prohibitively complex even for the small-scale topology, but the branch and bound (BB) algorithm searches the variable space more cleverly. However, the worst case computational complexity of any BB search algorithm is same as the complexity of the brute-force search in the large scale network, and this motivates the search for more efficient heuristics. In addition, the service fairness concept in multiuser multiservice International Mobile Telecommunication-Advanced (IMT-Advanced) Macro-Femto Networks has been defined, and the Jain’s fairness index has been modified. Compared to the traditional fairness index, the modified index is more proper to evaluate the service fairness of radio resource allocation algorithms for future multiservice wireless networks. Simulation results show that the proposed heuristic scheme can achieve system throughput very close to the upper bound of the optimal solution by BB algorithm in the moderate-scale network and higher than those of the simple cross-layer design (SCLD) scheme and the no CLD scheme in the large-scale network. In addition, the proposed heuristic algorithm can guarantee the QoS of different kinds of services, the fairness of various kinds of services, and user fairness.The power consumption minimization problem of Macro-Femto Networks has been formulated. The QoS requirements have been delivered to the physical layer, the MAC layer, the network layer and the transport layer. The constraints of power and RB allocation in the physical layer, delay and target data rate in the MAC layer, urgent queue length in the network layer, and packet error rate in the transport layer have been considered. After the restrictions of upper layers are translated into constraints with physical layer parameters by utilizing the mathematic theory, and the integer restrictions are relaxed, the original non-deterministic polynomial time hard (NP-hard) problem can be decomposed into convex optimization subproblems. The optimal solutions of RB allocation and power allocation can be obtained by using the Lagrangian optimization. Simulation results show that the proposed scheme is better than both the round robin algorithm and the max min one in terms of the energy efficiency, the throughput and the service fairness. The round robin algorithm and the max min one only focus on the user fairness rather than quality of service fairness. Compared to the round robin scheme (the max min one), the proposed scheme improves the energy efficiency58.85%(62.41%), the throughput19.09%(25.25%), the service fairness57.69%(35.48%).The throughput maximization problem based on cross-layer design has been formulated in the presence of imperfect channel state information (CSI). The outage probability imposed by the imperfect CSI and QoS provisions are considered. The dual decomposition method (DDM) is used to solve the problem. When the number of RBs is infinite, the duality gap is0. However, the number of RBs in the realistic network is a positive integer. Thus, an improved genetic algorithm with close to optimal performance and low complexity has been proposed. In our proposed cross-layer design-improved genetic algorithm (PCLD-IGA), if the similarity of two parent chromosomes exceeds the predefined threshold, mutate the parent with lower fitness, in order to avoid evolutionary retardation, and help better individual survive. What’s more, the PCLD-IGA develops the adaptive mutation probability. When the fitness of the user is smaller than the average fitness of the generation, the mutation probability is higher. Otherwise, the mutation probability decreases. The larger the fitness, the lower the mutation probability. Simulation results demonstrate that the PCLD-IGA with close to optimal performance outperforms the existing solutions in terms of system throughput, resilience to CSI errors and QoS performance.