Exploiting the Unicast Functionality of the On- Demand Multicast Routing Protocol Sung-Ju Lee, William Su, and Mario Gerla Wireless Adaptive Mobility Laboratory Computer Science Department University of California Los Angeles, CA

Unicasting using the Multicast protocol?  Generally not possible, or very inefficient  Most of the existing m-cast protocols (eg, AMRoute (Ad-hoc Multicast Routing), CAMP (Core Assisted Mesh Protocol), LAM (Lightweight Adaptive Multicast), etc) run on top of a SEPARATE unicast routing protocol  CAMP and LAM in particular, only work with certain underlying unicast protocol EXCEPTIONS:  Multicast AODV (Ad-hoc On-demand Distance Vector) uses routes obtained from unicast AODV  ODMRP (On-Demand Multicast Routing Protocol) can transparently function as both multicast and unicast

ODMRP Overview  Mesh topology  Forwarding group concept  On-demand route construction  Soft state multicast group maintenance  Unicast capability  Mobility prediction

On Demand Multicast Routing Protocol  Forwarding Group: All the nodes inside the “bubble” forward the M-cast packets via “restricted” flooding  Multicast Tree replaced by Multicast “Mesh” Topology  Flooding redundancy helps overcome displacements and fading  FG nodes selected by tracing shortest paths between M-cast members FG Forwarding Group

Route construction in ODMRP  Similar to other on-demand routing protocols  Consists of a query and a reply phase  A source periodically transmits Join Query packets when it has data to send  Join Query packets can carry data payload to eliminate route acquisition latency  Intermediate nodes forward the packet and set up path back to the source (backward learning)  The destination sends a Join Reply in response to a Join Query

Key Differences from Other On-Demand Protocols (e.g., DSR, AODV)  Intermediate nodes can not reply from cache  Data payload piggybacked on Join Queries must reach destinations  Routes replied by destination are more up to date  Query packets are periodically sent as long as there are data packets to send  Fresh routes are continually built and used  Route refresh interval should be carefully selected

Unicast enhancement: Mobility Prediction  Mobility prediction can help determine longevity of routes and schedule refresh requests  Mobility can be predicted, e.g., in an outdoor environment by means of GPS location information; received power based prediction also possible  Join Queries are flooded only before predicted route disconnection time  The scheme adapts refresh interval to mobility patterns and speeds

Route Selection Criteria at Destination  Route 1 is selected if the delay is the criterion  Route 2 is selected if the longevity is the criterion

Performance Evaluation  Simulated in GloMoSim written in PARSEC  Compared the performance of the following schemes:  ODMRP  ODMRP-MP: ODMRP with mobility prediction  WRP (Wireless Routing Protocol): an ad hoc distance vector routing protocol  LAR (Location Aided Routing): an on-demand protocol that uses GPS location information  50 nodes in 1000 meter X 1000 meter area  Free space propagation model, IEEE DCF  Random mobility model  Constant bit rate sources

Packet Delivery Ratio

ODMRP Packet Delivery Ratio as a function of refresh interval

Conclusions  ODMRP is capable of routing both unicast and multicast data effectively  Mobility prediction enhances ODMRP Unicast  Testbed implementation: presented at IEEE ICCCN 2000  Multicast work: ACM/Baltzer MONET special issue on multipoint communications 

The Number of Total Packets Transmitted per Data Packet Delivered

The Number of Control Bytes Transmitted per Data Byte Delivered by ODMRP with and without Mobility Prediction