Research On Scheduling And Interference Mitigation Techniques In LTE Systems

In order to satisfy people’s higher demand of data rate nowadays, the3rdGenerationPartnership Project (3GPP) released the Long Term Evolution (LTE) project, which aims at higherpeak data-rate, lower transmission delay, higher system capacity, and larger cell-coverage.Orthogonal Frequency Division Multiple Access (OFDMA), multiple-input and multiple-output(MIMO) techniques, scheduling and resource allocation schemes, etc., have been involved in LTEsystems to achieve its goal. However, in the premise of LTE standards, the system needs to deal withmany problems, such as how to exploit these key techniques efficiently, how to do user schedulingand resource allocation effectively, and in the uplink how to control the user’s transmit power, and soon. Moreover, although the intra-cell interference can be avoided effectively based on (orthogonalfrequency division multiplexing) OFDM, however, the inter-cell interference in practical multi-cellsystems still exists, which will severely limits the throughput of cell-edge users.This thesis does a specific and extensive study to deal with the complexities and challengesabove, and its novelty can be summarized in the following:1. User classification, user scheduling, and transmission modes selection are investigated inLTE downlink systems with both high-and low-mobility users. In LTE downlink, the basestation usually performs scheduling and resource allocation according to the channelinformation reported by users. For low-mobility users, frequency selective scheduling (FSS)algorithm can be used to select resources for them because of the accurate feedbackchannel information. However, for the high-mobility user, its channel changes very fast.The channel information reported in a specific transmission time interval (TTI) in LTEsystems is not very accurate. It will lead to performance loss for high-mobility users ifscheduling them based on the inaccurate channel information. Therefore, frequencydiversity scheduling (FDS) is usually used to assign high-mobility users frequencyresources, which are distributed in the frequency domain uniformly, to ensure the stableperformance gain. Thus, in this thesis, user classification to identify high-and low-mobilityusers, scheduling, and transmission mode selection under heterogeneous mobility scenariowill be studied firstly. More specifically,1) A scheduling algorithm considering frequency and multi-user diversity is proposed todeal with the problem that different scheduling and resource allocation schemes needto be used for the scenario with both high-and low-mobility users. It uses channelquality indicator (CQI) reported by users to classify users first, and then exploits FSSand FDS to assign resources to low-and high-mobility users, respectively. When classifying users, it first collects the reported CQI information within a given timesegment, computing the CQI variance, taking the average of CQI variances within thetime segment, and then compares with a given variance threshold to identify users.When scheduling users, it first determines the optimal low-mobility user for eachresource block (RB) based on FSS, then finds the optimal high-mobility user fordiversified RBs by FDS, and finally the optimal high-mobility user will overwrite theoptimal low-mobility one on the diversified RBs if it can bring performanceimprovement. The proposed algorithm ensures frequency-selectivity gain andfrequency-diversity gain for low-and high-mobility users, respectively, thusimproving the systems performance significantly. Moreover, the computationalcomplexity of the proposed algorithm is comparable with that of FSS; therefore, it isready for practical implementation.2) The channel delay spread affects the accuracy of user classification based on reportedCQI. First, a robust user classification algorithm is proposed to deal with this problem.The algorithm is based on the sign function. It takes the CQI variance which is largerthan a given threshold to be a constant while the one less than a given threshold to bezero, then computes the average variance value for different time segment, and finallygets the final output. Then, the proposed algorithm is extended to MIMO systems.According to the CQI computing scheme, it first converts the CQIs of different spatiallayers in the same RB to the effective signal-to-interference plus noise ratio (SINR),then deriving one effective SINR value of these SINRs, and finally computes theeffective CQI value and do user classification. Then, a low-complexity schedulingalgorithm is proposed. It schedules users based on PF metric. For high-mobility users,the computation of PF metric is based on the average data rate over the entire systembandwidth while for low-mobility users; it is based on the data rate on each RB. Theproposed scheduling algorithm guarantees frequency diversity gain of high-mobilityusers by using the average data rate and ensures frequency selectivity gain oflow-mobility users by exploiting the instant data rate. The proposed userclassification algorithm reduces the effect of channel delay spread on the CQIvariance computation value, and it is very easy and effective.3) A joint scheduling and transmission mode selection scheme for scenarios with bothhigh-and low-mobility users is proposed to deal with the problem that differenttransmission mode and scheduling algorithm will affect the system performance. Thealgorithm first requires that all the users transmit signals based on open-loop spatialmultiplexing (OLSM). The base station collects the CQI information reported by users and classifies users into high-and low-mobility ones. Then, OLSM will be usedby high-mobility users and closed-loop spatial multiplexing (CLSM) will be used bylow-mobility users, and users are scheduled based on (proportional fair) PFscheduling algorithm. When scheduling users, the computation of PF metric forhigh-mobility users is based on the average data rate across the entire bandwidthwhile the computation of PF metric for low-mobility users is based on the averagedata rate on each subband. The proposed algorithm selects the different transmissionmode and PF metric computation method for high-and low-mobility users, thus thesystem performance can be improved significantly.2. Multi-cell cooperative scheduling and power control to mitigate the interference in LTEuplink systems is investigated in this paper. In multi-cell systems, inter-cell interference is amajor factor that limits the system performance. In LTE downlink systems, the interferencecomes from base stations with constant physical positions. When equal power allocationscheme is used, the inter-cell interference is affected slightly by user scheduling andresource allocation. However, in the uplink, the interference is from different users.Because the physical position and transmit power is different among users, user schedulingand power control will affect the inter-cell interference significantly, and thus havingimpact on the system performance. Generally, multi-cell cooperation can mitigateinterference effectively. By sharing user’s channel state information (CSI), user schedulingand resource allocation information, and transmit power configuration information, user’sresources can be assigned and transmit power can be configured effectively, thus mitigatingthe inter-cell interference and improving the system performance. Therefore, this thesis alsostudies the cooperative scheduling and power control schemes for LTE uplink multi-cellsystems. More specifically,1) A cooperative scheduling algorithm is proposed to deal with the problem that theperformance of multi-cell systems deteriorates by inter-cell interference. Theproposed algorithm exploits the accurate inter-cell interference information of thescheduled cells and estimated interference information from other non-scheduledcells when scheduling users in a certain cell. It first divides all the cells in the networkinto several clusters. The cells within the same cluster share user’s channelinformation, user scheduling and resource allocation information and transmit powerinformation while the cells in the different clusters do not share any information.Then, the scheduling order of the cells in the same cluster is in the ascending order ofthe moving average throughput of each cell. The scheduling of each cell considersintra-cluster interference and inter-cluster interference and introduces weight factor to compensate the effect of user number differences among cells. The proposedalgorithm takes interference caused by user selection into account, thus mitigating theinterference effectively and improving the system performance significantly.2) A joint cooperative scheduling and power control algorithm for multi-cell systems isproposed to deal with the inter-cell interference in LTE uplink. It performs schedulingand power control for each cell separately, which schedules users first and thenconfigures the transmit power for each user. The proposed algorithm performsscheduling by first estimating the inter-cell interference and then assigning resourcesto users. When optimizing the user’s transmit power, both the performance variationof the objective cell and other interfering cells is considered. Furthermore, alow-complexity power control algorithm is proposed, which only considers theperformance change of several cells interfered most by the objective cell andestimates the performance change of all other cells by introducing a compensationfactor, when optimizing the transmit power of each user. Simulation resultsdemonstrate the effectiveness of the proposed algorithm in cell average and cell edgethroughput.