| Mobile-data traffic and multimedia applications have been growing explosively with the development of mobile internet. To face the ever growing quality of service (QoS) demand for mobile broadband systems, the3rd generation partnership project (3GPP) has introduced the long term evolution (LTE) as the next step of the current3rd generation telecommunication (3G) mobile networks. Consequently, it can have great significance in investigating all aspects of LTE system. LTE adopts multiple input multiple output (MIMO) and orthogonal frequency division multiplexing (OFDM) to improve its system performance. Due to the limitation of radio resource including time, frequency, power and so on, it is a great challenge to allocate the resources to different users effectively and it has always been a research focus. Lots of current research has been using cross-layer method to solve the resource allocation problem for all users. However, they cannot consider the QoS requirements in application layer, buffer state in radio link control layer, link state in physical layer at the same time.To allocate resources effectively in a time division long term evolution (TD-LTE) downlink system, a cross-layer resource allocation algorithm has been proposed which considers the QoS requirements in application layer, buffer status in radio link control layer, link state in physical layer and so on. In addition, we have established the four channel models in TD-LTE and analyze the performance of our channel models compared to the SMU200A of ROHDE&SCHWARZ (R&S). The main work in this thesis is included as follows:At first, we have summarized the merit and demerit of three traditional resource allocation algorithms in TD-LTE system including the Max C/I, Round Robin and Proportional Fairness. In addition, we have generalized the channel modeling methods of the large-scale fading and small-scale fading in a wireless channel.Then a joint resource allocation (JRA) algorithm with the objective of maximizing the overall system throughput while achieving fairness among different types of services is proposed. The scheme considers the constraints of diverse QoS requirements and several parameters which have great effect on the decision of MAC scheduling, including buffer state, subchannel link state, subchannel occupancy state. Our RA strategy solves the optimization problem with linear complexity and involves three stages. In stage one, we solve the convex optimization problem with Lagrange multiplier technique to find the recommended subchannel allocation matrix. In stage two, through the water-filling power allocation (WPA) algorithm, the optimal power allocation matrix is derived according to the recommended channel allocation matrix. In stage three, through an iterative algorithm based on the greedy algorithm, the subchannels and power are actually allocated to users considering their buffer states and QoS requirements according to the previous recommended subchannel and power allocation matrix. Simulation results have indicated that the proposed scheme has achieved the higher throughput and better QoS requirements than the traditional RA algorithms.At last, we have established the four channel models in TD-LTE including Extended Pedestrian A model (EPA), Extended Vehicular A model (EVA), Extended Typical Urban model (ETU) and High Speed Train scenario (HST). In addition, we have tested the channel models from two aspects. Firstly, the autocorrelation function and Doppler spectrum are simulated and tested. Then we analyze the performance of our channel models compared to the SMU200A of R&S in a link-level simulation platform. With the channel models, we can analyze the channel conditions in different scenarios and make appropriate adjustments in resource allocation algorithms. |
PDF Full Text Request