Algorithms for Routing and Centralized Scheduling in IEEE Mesh Networks Harish Shetiya and Vinod Sharma Department of Electrical Communication Engineering Indian Institute of Science WCNC 2006
Outline Introduction System Model Routing Scheduling Algorithm Simulation Conclusion
Introduction The IEEE d specifies a centralized scheduling scheme for mesh networks The MSSs notify the MBS Data transfer requirements The quality of their links to their neighbors The MBS uses the topology information along with the requirements of each MSS to decide the routing and the scheduling
System Model MSSs are labeled 1,2,3 …… M need not be directly connected to the MBS MBS is labeled by 0 Each node has infinite buffer to store the packets
Routing 1 M 2 BS a frames consisting of N time slots I 點的資料經過 J 點的 平均速率 P 2,1 P 2,0
Scheduling Algorithm Tree Dynamic Programming Framework Maximum Transmission Scheme Per Slot Maximum Transmission Scheme Maximum Work Scheme Fixed Allocation Scheme Adaptive Fixed Allocation Scheme Order scheme Downlink Scheduling Algorithm
Tree Dynamic Programming Framework_ Tree Structure BS A frame has N slots
Tree Dynamic Programming Framework_ Uplink Scheduling denote the maximum data extracted from node i in k slots denote the maximum reward obtained from node i in k slots
Tree Dynamic Programming Framework_ Maximum Transmission Scheme Construct a table : (1)First get: (2)then get: A B C rbrb rara rcrc
Tree Dynamic Programming Framework_ Example (schedule b, c) BS A A frame has N slots, One slot 1 s B C 6 M 2 M/S 2 M 1 M/S 3 2 k 值 {node} N {} {B} {B,C} (1,0)(2,0)(3,0)(3,1)(3,2)
Tree Dynamic Programming Framework _ Example (schedule a, b, c) Construct a table : B,c 在 (k-n) 個 slots 所傳的資料 量 min{A 點 queue 中的資料量 +(b,c 傳進的資料量 ), A 能傳送的資料量 } A B C rbrb rara rcrc
Tree Dynamic Programming Framework_ Example (schedule a, b, c) A frame has N slots, One slot 1 s k 值 {node} N {B,C} 0 (0,0) 2 (1,0) 4 (2,0) 6 (3,0) 7 (3,1) 8 (3,2) 8 {A,B C} 0 (0,0) 4 (1,0) 8 (2,0) 10 (2,1) 12 (2,2) 16 (3,2) BS A B C 6 M 2 M/S 2 M 1 M/S 3 2 8M 4M/S
Tree Dynamic Programming Framework_ Example (schedule b, c) A frame has N slots, One slot 8 s BS A B C 6 M 2 M/S 8 M 1 M/S k 值 {node} N {B, C} 0 (0,0) 8 (0,1) 14 (1,1) 14
Tree Dynamic Programming Framework_ Per Slot Maximum Scheme Suboptimal version of Maximum transmission scheme i 點的傳 輸 rate i 點的 queue 資 料量 A3M4M/s B5M2M/s C25M/S
Tree Dynamic Programming Framework_ Maximum Work Scheme Total work in the system at the beginning of the jth frame (work is the time to extract entire data from the network ): 1 M 2 BS P 2,1 P 2,0
Tree Dynamic Programming Framework_ Maximum Work Scheme b,c 所需 的時間 a 所須的 時間 A B C rbrb rara rcrc 7 M 3 M/S 1M 1M/S 1 3 Slot Full Slot
Tree Dynamic Programming Framework_ Fixed Allocation Scheme 其他點經過 I 的 資料量
Tree Dynamic Programming Framework_ Adaptive Fixed Allocation Scheme(1) denote the set of good and bad links Ex. First round: is the upper bound of accumulated credit
Tree Dynamic Programming Framework_ Adaptive Fixed Allocation Scheme(2) second round: Slots to assign good link
Tree Dynamic Programming Framework_ Order Scheme To get the order: The slot is assigned to node :, i has higher rate of data delivery to MBS + E點:E點: ++ = 1
Tree Dynamic Programming Framework_ Downlink Scheduling Algorithm there are M possible destinations in the downlink. Hence we need to track the data as it passes through the network. Since all the traffic originates at the MBS. Hence, using schemes for the downlink in a manner that is analogous to the uplink.
Simulation Mesh network of 10 nodes
Simulation_ Physical Layer Parameter
Simulation_ Burst Profile
Simulation_ comparison of uplink scheduling scheme
Simulation_ Comparison between different Fixed Allocation Scheme
Simulation_ Comparison between different Maximum Transmission Scheme
Simulation_ Comparison of downlink scheduling scheme
Conclusion To develop efficient routing and scheduling algorithm for multihop wireless networks as formed by IEEE standard