MPR-Aware Performance Improvement for Inter-Domain Routing in MANETs You Lu, Yenan Lin, Biao Zhou, Mario Gerla UCLA
Introduction & Previous Work New features Outline Introduction & Previous Work New features Dynamic TDMA MPR-aware Slot Allocation for OLSR MPR-aware Gateway selection Simulation Results Throughput increased efficiency Reduced Control Overhead Conclusions
Inter-Domain Routing for MANETs Previous Work Inter-Domain Routing for MANETs Group Mobility One Cluster head per Group Cluster heads periodically advertise member list Advertisements propagate via dissemination A LISTA B LISTB
TDMA domain is the bottleneck Previous Work TDMA domain is the bottleneck Inefficient slot usage Three categories of TDMA Static TDMA Cluster TDMA Dynamic TDMA
Our Approach: DTDMA Slot Allocation for Node 1 contention area
DTDMA performance much better than STDMA DTDMA with OLSR +MPR DTDMA performance much better than STDMA Same slot reused in different contention areas However: Can we further improve performance? Yes, if OLSR is used in the TDMA domain Key idea: assign multiple slots to “popular” nodes, such as Multipoint Relay Nodes (MPR)
OLSR: Multi Point Relays
New Feature: multiple slots for MPRs MPR-aware slot allocation: MPR nodes get more slots than regular nodes Motivation: MPRs form the “backbone” They are the nodes that do most store and forwarding MPRS = MPR Selector, the node that selects that MPR The more Selectors, the more popular the MPR
DTDMA : MPR Selectors Selectors for MPR in node 3 MPRS MPRS MPR MPRS
MPR- aware Gateway selection More on MPR MPR- aware Gateway selection - Can further improve performance by selecting as Gateway a popular MPR Node - Assign candidate Gateways the WEIGHT:
Performance Evaluation Simulator: Qualnet 5.0.2 Data traffic: CBR, pkt size 512 byte Geographic area: 2000m × 2000m Mobility model: RPGM In each domain the Reference Point moves randomly. All nodes in the domain move around the RP intra-group speed: 0-5m/s pause time is 10 seconds MAC protocols: CSMA & TDMA Routing protocols: AODV & OLSR Metrics: Average Receiver Throughput & Normalized Control Overhead
Average Delivered Throughput The larger the density, the larger the frame for the Contention set => the lower the throughput
Normalized Control Overhead
Conclusion Dynamic TDMA +OLSR: We get: MPR-aware Slot Allocation MPR-aware Gateway Selection We get: Throughput improvement Reduced Control Overhead
Thanks Q & A