Measurementbased Admission Control for multihop Wireless Mesh Networks

Measurement-based Admission Control for multi-hop Wireless Mesh Networks Dhruv Gupta EEC 273 class project Prof. Chen-Nee Chuah 2/21/2021 Dhruv Gupta, EEC 273 1

Wireless Mesh Networks 2/21/2021 Dhruv Gupta, EEC 273 2

Admission Control Ø Ø Ø Is the simple practice of discriminating which traffic is admitted into a network The network judges whether it has enough resources available to accept a new connection based on the current network status Used to control the amount of traffic in the network and thus provide guaranteed service to end users 2/21/2021 Dhruv Gupta, EEC 273 3

Background n WMN’s have become increasingly prevalent ¨ Roofnet, SF, Mountain View, Philadelphia and many others n Ease of deployment, increased coverage area and reduced cost have driven the deployment of WMNs n At the same time, low bandwidth, interference and high loss rates are some of the problems n Network management is of crucial importance to provide guaranteed service to end users 2/21/2021 Dhruv Gupta, EEC 273 4

YAO: Yet Another Outline n Problem statement n Related work n Motivation n Methodology n Conclusion 2/21/2021 Dhruv Gupta, EEC 273 5

Problem Definition n I/P : Connectivity graph & measurement data Objective : Perform an AC decision for a given client request by finding a path P from source to destination that satisfies the following constraints Constraints : For each link ‘i’ on path P, we have Buser <= α * ABi 2/21/2021 Dhruv Gupta, EEC 273 6

Related work n n Yaling Yang, Robin Kravets, "Contention-Aware Admission Control for Ad Hoc Networks, " IEEE Transactions on Mobile Computing , vol. 04, no. 4, pp. 363 -377, July/August, 2005. Luo, L. , Gruteser, M. , Liu, H. , Raychaudhuri, D. , Huang, K. , and Chen, S. 2006, “A Qo. S routing and admission control scheme for 802. 11 ad hoc networks, ” In Proceedings of the 2006 Workshop on Dependability Issues in Wireless Ad Hoc Networks and Sensor Networks (Los Angeles, CA, USA, September 26 - 29, 2006), DIWANS '06 2/21/2021 Dhruv Gupta, EEC 273 7

Motivation n Most schemes focus on wireless LANs or Adhoc networks Simulation based studies do not capture the underlying interference models accurately No experimental measurement based work yet for wireless mesh networks 2/21/2021 Dhruv Gupta, EEC 273 8

Assumption n n Small scale WMNs with single gateway node Single channel for wireless backhaul and separate channels for clients All traffic directed for the internet – no peer to peer communication within the network Client based admission control, not flow-based User resource requirements are assumed to be known 2/21/2021 Dhruv Gupta, EEC 273 9

Problem components… n Connectivity graph n Network measurements n Centralized admission control algorithm 2/21/2021 Dhruv Gupta, EEC 273 10

Network Discovery n Purpose: ¨ Advertise node existence/nonexistence ¨ Construct connectivity graph n Constraint: ¨ Achieve node advertisement with minimum resource usage ¨ Accurate record of neighboring nodes 2/21/2021 Dhruv Gupta, EEC 273 11

Hello Message - Types n Add - new node being added n Delete - node presumed dead n Update - “I’m alive” 2/21/2021 Dhruv Gupta, EEC 273 12

Network measurement n Why? ¨ Centralized controller needs an updated view of the network topology statistics n What? ¨ Bandwidth of the links in the network (asymmetric) 2/21/2021 Dhruv Gupta, EEC 273 13

How to get Link Bandwidth? n n n A wireless link is different from a wired link Wireless is a broadcast medium, which means traffic on one link will affect bandwidth on another What is the base variable for bandwidth? ¨ Amount 2/21/2021 of Data/Time Dhruv Gupta, EEC 273 14

Time = Channel Utilization n Channel Utilization is used instead of node utilization because CU is the limiting factor (and is affected by other APs) We define channel busy time as the total time a node is busy transmitting, receiving or sensing packets Channel utilization = CBT / tmeasurement 2/21/2021 Dhruv Gupta, EEC 273 15

Amount of Data = Modulation Rate n n The maximum bandwidth allowed per link (src to dest) varies based on signal quality and packet loss rate General upper bound for amount of data that can be sent to a certain neighbor 2/21/2021 Dhruv Gupta, EEC 273 16

Thus, n n Upper bound bandwidth to a neighbor = Modulation Rate * (1 - Channel Utilization) Monitored data is periodically sent to the central controller 2/21/2021 Dhruv Gupta, EEC 273 17

The final piece n Central Controller n Implements the admission control algorithm n Takes as input: § Connectivity graph § Available Bandwidth 2/21/2021 Dhruv Gupta, EEC 273 18

Path search algorithm n Use a uniform cost search algorithm to search for a path From source node, sort the links in decreasing order of available bandwidths n Check if first link satisfies the user requirement n § If not, then request cannot be accepted § Else repeat the procedure for subsequent links 2/21/2021 Dhruv Gupta, EEC 273 19

Centralized Admission Control n Client associates to an AP (Wi-Fi assoc req) n AP will send a client association request to centralized controller n Centralized controller will reply (after running admission control algorithm) with a client association reply n AP will receive the reply and send a Wi-Fi assoc reply back to the client n If a path is found, send back association reply and setup route n If no path is found, send back a reply denying the new request n If a new client is accepted or a client leaves the network, the measurements will reflect the same 2/21/2021 Dhruv Gupta, EEC 273 20

Implemented the network discovery protocol ü Implemented the central controller algorithm Ø Currently working on estimation of available bandwidth Ø Integrate and test the scheme ü 2/21/2021 Dhruv Gupta, EEC 273 21

n Thank You n Hope you all get A’s and I get an A+ n Q&A 2/21/2021 Dhruv Gupta, EEC 273 22