GPSR: Greedy Perimeter

Stateless Routing for Wireless

Networks
EILEEN BALCI

About GPSR
GPSR allows nodes to figure out who its closest

neighbors are (using beacons) that are also close to

the final destination the information is supposed to
travel to
To calculate a path, GPSR uses a greedy
forwarding algorithm that will send the
information to the final destination using the most
efficient path possible. If the greedy forwarding
fails, perimeter forwarding will be used which
routes around the perimeter of the region.

About the Greedy Forwarding Algorithm

Assuming the wireless routers [nodes] know their

own locations the Greedy forwarding algorithm will

try to find the closest router which is also the closest
to the final destination

Greedy Forwarding Algorithm Benefits

A node just has to remember the location of

neighbors with in one-hop

Routing decisions can be dynamically made

Greedy Forwarding Algorithm Drawback

If the network is dense enough that each interior

node has a neighbor in every 2/3 angular sector,

Greedy Forwarding will always succeed. However,
the greedy forwarding algorithm can fail:
Greedy Forwarding fails

Perimeter Forwarding Algorithm Benefits

When the Greedy Forwarding algorithm fails, the

Perimeter Forwarding algorithm will be used

Apply the right-hand rule to traverse the edges of the

void and find a path using the topologys perimeter

Perimeter Forwarding Algorithm Drawback

The Perimeter Forwarding Algorithm uses a longer

path to the destination so the perimeter forwarding

algorithm less efficient and cannot be used alone

Greedy Perimeter Stateless Routing - GPSR

Putting Greedy Forwarding and Perimeter

Forwarding together makes the final GPSR which

will use the necessary algorithm(s) to find the best
path in a given topology
greedy fails
Greedy Forwarding

Perimeter Forwarding

have left local maxima

greedy works

greedy fails

GPSR
The GPSR is a responsive and efficient routing

protocol for mobile, wireless networks. GPSR can be

applied to Sensor networks, Rooftop networks,
Vehicular networks and ad-hoc networks.

Credits
www.cs.binghamton.edu/~kang/teaching/cs580s/g

psr.ppt
http://www.eecs.harvard.edu/~htk/publication/2000-mobi-karp-ku

ng.pdf
http://www.icir.org/bkarp/gpsr/gpsr.html