Download presentation
Presentation is loading. Please wait.
Published byOphelia Fields Modified over 8 years ago
1
1 Guard Channel CAC Algorithm For High Altitude Platform Networks Dung D. LUONG TRAN Minh Phuong Anh Tien V. Do
2
2 Document Properties
3
3 Outline Call Admission Control (CAC) in HAP-based networks (the work done within the framework of the European CAPANINA project) The use of the guard channel algorithm in CAC for HAP-based networks to provision multiple traffic classes Computational problem Direct solution vs. proposed approximation (a new algorithm) Validation results Conclusions
4
4 HAP-based network scenario
5
5 HAP cell BS New call admitted New call blocked Blocking Batch arrival Handoff call admitted Handoff call dropped Dropping HAP cell
6
6 Channel Guard CAC New call Admitted Blocked Handoff call Admitted Dropped Total bandwidth (C) Channel guard (G) Existing traffic New call #1 time bandwidth New call #2Handoff call #1Handoff call #2
7
7 Channel guard for traffic classes Traffic classes Class 1 Class 2 Class 3 Class 4 bandwidth 0 (max. bandwidth) C G4G4 G3G3 G2G2 G1G1
8
8 Problem description Given Service classes with their parameters (bandwidth requirement, arrival rate, departure rate) Capacity of the system Objective: how to find optimal channel guards, so that Dropping probability is below a certain threshold Maximalize the bandwidth usage The key is To have an efficient method to calculate the dropping probability To have an algorithm to find the G i value
9
9 Calculation of dropping probability Two approaches can be applied Direct solution of the multi-dimensional Markov chain (time consuming) Approximation Equivalent traffic Build up an one-dimensional Markov-chain steady state distribution Find dropping probability for each state Calculate dropping probability
10
10 Multi-dimensional Markov chain 0,N 2 0,N 2 -10,10,02,01,11,02,11,N 2 1,N 2 -12,N 2 2,N 2 -1N 1 -1,0N 1 -1,1N 1 -1,N 2 N 1 -1,N 2 -1N 1,0N 1,1N 1,N 2 N 1,N 2 -1 …… … …
11
11 Calculation of dropping probability (2) Equivalent traffic Build up a Markov-chain steady state distribution Find dropping probability for each state Calculate dropping probability
12
12 Calculation of dropping probability (3) Equivalent traffic Build up a Markov-chain steady state distribution Find dropping probability for each state Calculate dropping probability 012HiHi H i +1 ……
13
13 Calculation of dropping probability (4) Equivalent traffic Build up a Markov-chain steady state distribution Find dropping probability for each state Calculate dropping probability Remaining bandwidth exp. val. of bandwidth requirement of a batch
14
14 Calculation of dropping probability (5) Equivalent traffic Build up a Markov-chain steady state distribution Find dropping probability for each state Calculate dropping probability Dropping probability Steady state distribution Conditional dropping probability, when there are j calls in the system
15
15 Calculation of the Channel Guard vector Search between possible channel guard values For all searched value, calculate dropping probability Find the largest value to ensure maximum bandwidth usage Example: dropping threshold = 1% Linear searchBinary search 0.30.10.050.0090.0020.050.009 80797877 channel guard dropping probability channel guard 757877
16
16 Validation Dropping threshold Simulation G 1 Simulation G 2 Model G 1 Model G 2 0.090.90.950.90910.9075 0.080.90.950.90370.9006 0.070.90.950.89840.8932 0.060.950.90.89300.8897 0.050.950.90.88240.8790 0.040.950.90.87170.8683 0.030.9 0.86100.8577 0.020.9 0.83960.8399 0.010.850.90.80750.8078
17
17 Conclusions The application of guard channel algorithms for HAP- based networks We proposed a fast method To approximate optimal channel guards Takes batch hand-offs into account Future work Calculate approximation errors Calculate confidence intervals
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.