Global State Routing (GSR) maintains a global knowledge of network topology like link state routing but avoids inefficient flooding. Each node periodically exchanges its link state table only with neighbors, not the entire network. This reduces overhead compared to link state routing. GSR nodes use Dijkstra's algorithm on the accumulated link state information to compute optimal paths locally without global flooding.

1. 1
Global State Routing
• Global State Routing (GSR) is based on Link State (LS)
routing. In the LS routing method, each node floods
the link state information directly into the whole
network (global flooding) once a link change
between itself and its neighbors is detected. A node
gets to know the whole topology by obtaining link
information. LS routing works well in static topology
networks. If links change quickly at high mobility,
frequent global flooding will lead to huge control
overhead (large amount of small packets).

2. 2
Global State Routing
• Aim:
The knowledge of full network topology as LS routing should be
maintained, but the inefficient flooding mechanism has to be
avoided. Unlike LS, GSR does not flood link state packets.
Instead, every node maintains its link state table based up-to-
date( LS information received from neighboring nodes) It will
periodically exchange its LS information with its neighbors only
(no global flooding). This means that GSR is MAC (medium
access control) layer efficient as it keeps the overhead of
control message low. GSR still finds accurate and optimal paths.
GSR could be described as being based on LS routing, which has
the advantage of routing accuracy, and the dissemination
method used in DBF, to avoid inefficient flooding like in LS
routing.

3. 3
Global State Routing
• Each node maintains:
• a neighbor list containing the list of nodes adjacent to the node ( hop=1 )
• a topology table containing the link state information reported by a
destination and a timestamp indicating the time at which this has been
reported.
• a next hop table containing the next hop to which the packets for this
destination have to be forwarded a distance table containing the shortest
distance to each destination node Initially, each node learns about its
neighbors by examining each received packet
and thus builds up its neighbor list Each node updates link state
information in its topology table by receiving link state messages from its
neighbors. LS packets with larger sequence numbers replace the older ones
with smaller sequence numbers. So every node learns the entire network
topology. The entire topology map (link state table) is exchanged
periodically with neighbors
only, meaning that there is no global flooding. Then each node computes
the shortest paths itself using the newly rebuild topology
map, based on Dijkstra’s algorithm.
• In summary this means that based on the link state vectors, nodes maintain
a global
knowledge of the network topology and take their routing decisions
locally.

4. 4
Global State Routing
• Advantages
GSR greatly reduces the control overhead as it avoids flooding for
disconnects/reconnects and updates are time triggered than event
triggered. The routing accuracy of GSR is comparable to an ideal LS
scheme and thus superior to the traditional DBF. A bandwidth function
can be used to realize QoS routing.
• Disadvantages
The main disadvantage is the large size of the routing message.
• As the entire topology table
is broadcasted with each update, a considerable amount of bandwidth is
consumed.
• The latency
of the link state change propagation depends on the update period,
meaning that it has to be
carefully chosen.