EECS 600 Adavanced Computer Networks

Notes Date: February 21, 2005

Paper: A scalable Location Service for Geographic Ad Hoc Routing

1. Assumption for the routing algorithm proposed

All nodes agree on the global origin of the grid hierarchy.At startup, all the nodes know the same global partitioning of the world into a hierachy of grids of increasing size.

2. How does a node selects its location servers?

A node selects location servers in each sibling of a square that contains the node.In other words, the node selects three location servers for each level of the grid hierarchy.

3. What is the advantage of the selection technique proposed in above point.

A node will have more location servers in its vicinity and as the we go further away from the node, the number of location servers decreases. This will help the nodes which are very near to the target node since they can find more location servers for the target node.

4. what nodes are selected as location servers for a given node?

Every node will have an unique identifier. In each level, the node with least ID greater than that of the given node is recruited as location server.

5. What is the advantage of the above method?

This method spreads the load evenly over all the nodes. For example, the location servers for a node with ID 30 will have their IDs above 30. So, only the nodes with ID greater than 30 will keep the information for node with ID 30. Hence this technique shares the burden uniformly.

6. What are the main features of GLS (Grid Location Service)?

a) This is a decentralised service
b) Requires no fixed infrastructure
c) Tolerates node failures
d) It uses geographic forwarding
d) Only a small set of nodes are selected as location servers and are periodically updated.

7. Not more than n location query steps are needed to reah the destination when the source and destination are situated in an order-n square.

8. Grid's queries cause less network node since the queries are not flooded.This is the main difference between DSR and GLS.

9. Why does the query success rate decreases as the number of nodes in the space increases?

Though the density increases, the relative mobility of the nodes in a packet space will be high when compared to sparse space. Hence, the information maintained at intermediate servers will become out-of-date quickly.