考慮服務品質限制之具最大比率合成能力 中繼站無線網路成本最小化建置與路由策略 指導教授:林永松 博士 祝國忠 博士 研究生:劉翊恆 指導教授:林永松 博士 祝國忠 博士 研究生:劉翊恆 Minimum-Cost QoS-Constrained Deployment and Routing Policies for Wireless Relay Networks of Maximal Ratio Combining Capacities 國立台灣大學資訊管理研究所 碩士論文口試審查
2 Outline Introduction Problem Description and Formulation Solution Approach Computational Experiments Conclusion and Future Work
Introduction Background - Relay - IEEE j - Diversity Techniques - Maximum Ratio Combining (MRC) Motivation Background - Relay - IEEE j - Diversity Techniques - Maximum Ratio Combining (MRC) Motivation
4 Background – Relay Introduction Left: tree topology in relay network; Right: mesh topology in mesh network Relay technologies has been used widely in wireless communications, such as IEEE j, IEEE s, and seed concept in 3GPP Advantages of relay: - radio range extension - overcome shadow fading - reduce infrastructure deployment costs - enhance capacity - reduce outage probability
5 City scenario of relays deployment with one BS Background – Relay (Cont ’ d) Relays are designed to improve the coverage of a BS and overcome the shadows caused by obstacles. Three types of relay protocols: - Amplify-and-Forward : Relays act as analog amplifier. - Decode-and-Forward: Relays act as a digital repeater with the same codewords. - Decode-and-Reencode: Relays act as a digital repeater with different codewords. Introduction
Background – IEEE j IEEE j is now a developing specification (renamed from MMR, MMR stands for Mobile Multihop Relay) established by IEEE j task group The enhancement of original /802.16e-2005 Compatible to the legacy standard A relay station (RS) will be recognized as a base station (BS) by the mobile station (MSs) for the transparency reasons 6 Introduction
Background – IEEE j (Cont’d) 7 Introduction
Background – Diversity Techniques Frequency diversity: Transmitting or receiving the signal at different frequencies Time diversity: Transmitting or receiving the signal at different times Space diversity: Transmitting or receiving the signal at different locations Polarization diversity: Transmitting or receiving the signal with different polarizations 8 Cooperative diversity is a relatively new class of spatial diversity techniques that is enabled by relaying To improve the reliability of communications in terms of, for example, outage probability, or symbol-or bit-error probability, for a given transmission rate Introduction
Background – MRC Three major diversity signal- processing techniques: - selection diversity (SD) - equal gain combining (EGC) - maximal ratio combining (MRC) 9 Soft handoff Introduction
Motivation Base Station (BS) Relay Station (RS) Mobile Cluster (MC) 10 Introduction BS coverage cell coverage Inner zone: MC connects to the BS directly Outer zone: MC connects to the BS through RSs
Motivation (Cont’d) Allow multiple source nodes jointly transmit one single information if the signal strength is not robust enough in the links between one source node to the destination. To develop a wireless network topology based on j relay environment: - Where to build a RS and its configuration ? - Which RSs should a MC rout to ? - What is the routing policy between a BS and a MC ? The routing policy is no longer a single path but with more complex multicast-tree algorithms. 11 Introduction
Problem Description and Formulation Problem Description Problem Notation Problem Formulation Problem Description Problem Notation Problem Formulation
Problem Description 13 Problem Description and Formulation Base Station (BS) Relay Station (RS) Mobile Cluster (MC) Empty Location
Problem Description (Cont’d) Assumption: The relaying protocol in this model is Decode-and-Forward. Each MC must home to either a BS or relay(s). The relays selected by one MC must associate with the same BS. The routing path of each OD pair in DL (UL) is a multicast tree. The spatial diversity gains are represented by the aggregate SNRs with MRC techniques. The BER of a transmission is measured by the receiving SNR value. The aggregate BER of the destination are the summation of BER of each node on the routing tree. The numbers of links of each path adopted by each MC are assumed to be equal to ensure the MRC is achievable within limited delay. Error corrections and retransmissions are not considered in this problem. 14 Problem Description and Formulation
Problem Description (Cont’d) Given: The set of BSs, candidate RS locations, relay configurations, MCs Required data rate of a MC in DL and UL Fixed and configuration cost of a relay Distance between every two node Attenuation function Link SNR function The minimum SNR requirement for a MC in DL and UL to home to a BS or relay Link BER function The maximum BER threshold of a OD pair transmission in DL and UL Nodal and link capacity functions The maximum spatial diversity of a mobile cluster in DL and UL 15 Problem Description and Formulation
Problem Description (Cont’d) Objective: To minimize the total cost of wireless relay network deployment Subject to: Relay selection constraints Nodal capacity constraints Cooperative relaying constraints in DL and UL Routing constraints in DL and UL Link capacity constraints in DL and UL To determine: Whether or not a location should be selected to build a relay The cooperative RSs of each MC The DL and UL multicast tree of each MC 16 Problem Description and Formulation
17 Problem Notation Problem Description and Formulation
Problem Notation (Cont’d) 18 Problem Description and Formulation
Problem Notation (Cont’d) 19 Problem Description and Formulation
Problem Notation (Cont’d) 20 Problem Description and Formulation
Problem Notation (Cont’d) 21 Problem Description and Formulation
Problem Notation (Cont’d) 22 Problem Description and Formulation
Problem Notation (Cont’d) 23 Problem Description and Formulation
Problem Formulation 24 (IP 1) Subject to: (General Constraint) Objective function: Problem Description and Formulation
Problem Formulation (Cont’d) 25 Problem Description and Formulation
Problem Formulation (Cont’d) 26 Problem Description and Formulation
Problem Formulation (Cont’d) 27 Problem Description and Formulation
Problem Formulation (Cont’d) 28 Problem Description and Formulation
Problem Formulation (Cont’d) 29 Problem Description and Formulation
Solution Approaches Lagrangean Relaxation Method Problem Decomposition Getting Primal Feasible Solutions Lagrangean Relaxation Method Problem Decomposition Getting Primal Feasible Solutions
Lagrangean Relaxation 31 Lagrangean Relaxation Problem (LR) Primal Problem (P) Subproblem 1Subproblem 7 Optimal Solution LB <= Optimal Objective Function Value <= UB LB UB Lagrangean Dual Problem Adjust Lagrangean Multiplier Solution Approaches
Problem Decomposition 32 Subproblem 1 can be further decomposed into |R| independent problem. Time complexity: Solution Approaches
Problem Decomposition (Cont’d) 33 Subproblem 2 can be further decomposed into |R| x |B| x |DIR| independent problem. Time complexity: Solution Approaches
Problem Decomposition (Cont’d) 34 Subproblem 3 can be further decomposed into |N| independent problem to choose whether BS or RSs should MC n route to and the correlative SNR value. Time complexity: Solution Approaches
Problem Decomposition (Cont’d) 35 Solution Approaches Subproblem 4 can be further decomposed into |N| x |R| x |B| x |DIR| independent shortest path problem which can be optimally solved by bellman ford’s minimum cost shortest path algorithm. Time complexity:
Problem Decomposition (Cont’d) 36 Solution Approaches Subproblem 5 can be further decomposed into |R| x |R| independent problem to determine whether link uv be selected by MC n in DL and UL and the correlative SNR value. Time complexity:
Problem Decomposition (Cont’d) 37 Subproblem 6 can be further decomposed into |N| independent problem to determine the SNR value received by MC n in DL. Time complexity: Solution Approaches
Problem Decomposition (Cont’d) 38 Solution Approaches Subproblem 7 can be further decomposed into |N| x |R| independent problem to determine the SNR value received by RS v in UL transmission of MC n. Time complexity:
Getting Primal Feasible Solutions 39 Solution Approaches BS radius for MC BS radius for RS BER is over! SNR is not enough! X BS capacity is full X RS s which MC n routes to link uv which MC n selects Step 1: All RSs and MCs home to proper BS and sorted by the distances to the BS Step 2: Determine whether the BS or which RS should MC n route to refer to the coefficient of, then build the RS. Step 3: Find a shortest path from the selected RS to the BS via all built RSs with cost=BER of each link. Step 4: If the SNR of one link uv is not strong enough, find a shortest BER path between u and v with references of Step 5: If the BER value of the path is over the predefined threshold, repeat step 2 to step 4 to find another RS and path until the BER value is small enough. CheckCapacityofNode(); CheckCapasityofLink(); CheckLinkAmount(); SetConfiguration(); CheckCapacityofNode(); CheckCapasityofLink(); CheckLinkAmount(); SetConfiguration();
Computational Experiments Experiment Environments Experiment Designs Experiment Results Experiment Environments Experiment Designs Experiment Results
Experiment Environments Environment Parameters 41 Computational Experiments ParametersValueParameters Operation Frequency2500MHzAttenuation Factor3.2 Channel Bandwidth10MHzThermal Noise figure-174 dB BS Antenna Gain15dBiMin. RS to RS SNR dB RS basic Antenna Gain5dBiMin. SNR received by MC dB MS Antenna Gain-1dBiBER threshold BS noise figure4dBMax. Spatial Diversity3 RS noise figure5dBTraffic Required by MC (DL)1 Mbps MC noise figure7dBTraffic Required by MC (UL)0.5 Mbps BS Transmit Power43dBmBS Capacity100 M bps RS Basic Transmit Power33dBmRS Basic Capacity15 Mbps MC Transmit Power23dBmRS Fixed Cost1M dollars RS Config. Cost0.2M dollars From: “Mobile WiMAX”, WiMAX Forum, May 2006 Shadow Urban Area
Experiment Environments (Cont’d) Modulation and Code Rate 42 Computational Experiments ModulationCode RateSNRDL Rate (Mbps)UL Rate (Mbps) QPSK1/2 CTCSNR <= /4 CTC9.4 < SNR <= QAM1/2 CTC11.2 < SNR <= /4 CTC16.4 < SNR <= QAM2/3 CTC18.2 < SNR <= /4 CTC22.7 < SNR From: “Mobile WiMAX”, WiMAX Forum, May 2006
Experiment Environments (Cont’d) SNR Formulation: Path Loss Function: Thermal Noise Function:, transfer into (dB): MHz 10 MHz Noise Figure n: Attenuation Factor Computational Experiments Distance Transmit Power Transmit Gain Receive Gain
Experiment Environments (Cont’d) BER Function: 44 Computational Experiments
We proposed two topologies, grid and random, to compose the RS candidate locations, and examine two sizes of network radius with matrix of different number of RSs and MCs within one BS coverage. Then we proposed random topology with three different network radiuses within two BSs coverage to examine multiple BSs network environment. We introduced two algorithms to compare with the LR result: - Minimum BER Algorithm (MBA) - Density Based Algorithm (DBA) Experiment Designs 45 Computational Experiments
Experiment Results 46
Conclusion and Future Work Conclusion Contribution Future Work Conclusion Contribution Future Work
Conclusion Fixed MC number RS number increased=> Reduce cost Fixed RS number MC number increased=> Induce cost For a given networkscale, the farthest locations from BS to receive signals under BER threshold should be included in the candidate RS locations to reach the minimum cost objective. 48 Conclusion and Future Work
Contribution Constructed the network architecture with multicast tree routing concepts based on IEEE j specifications and spatial diversity techniques. Mathematically modeled the network development problem of previous environment. Proposed the solution approaches for engineering guidelines of RS buildings to minimize the total development cost. 49 Conclusion and Future Work
Future Work Applying different diversity techniques ex. Time diversity, frequency diversity...etc. Applying different fading models ex. Flat fading (time dispersion), Fast fading (doppler spread)...etc. Considering different performance matrixes ex. delay, throughput...etc. 50 Conclusion and Future Work
Thanks for Your Listening
MRC Verification 52