CCN Activity Session for Section 4-1 – MANETs
Today’s activities are on the topic of Mobile Ad-hoc Networks (MANETs).
Discussion
Why do reactive routing protocols have greater advantages in ad-hoc networks than in other classes of network? What properties of reactive protocols make them more suited to this environment?
Proactive routing protocols periodically distribute routing information between nodes, examples are optimised forms of link-state and distance-vector algorithms. Reactive protocols find routes on demand by flooding the network with discovery packets, examples include AODV and DSR. Ad-hoc networks are frequently used where node mobility is high and possible routes are more dynamic, as such there is less of an advantage to caching routes (as they become stale) and route discovery at each transmission is increasingly preferable.
Question
Why is flooding a comparably efficient strategy for control packets within a MANET, despite having a large overhead if used in a general routing context?
Control packets are small, and so it is often preferable to flood the network with small packets in order to ensure that frequent retransmission of much larger data packets is not required.
Discussion
What are the issues around broadcast storms? Why are they a particularly serious issue in MANETs? What are the potential mitigations?
Broadcast storms are the result of propagation down multiple paths where propagation delays are similar – they result in higher probability of collisions and message duplication. They are a particular problem in MANETs because of the extensive use of flooding in a local area.
Potential mitigations include:
• Probabilistic scheme – only re-broadcast the packet with a probability p (risks the
packet not reaching the destination)
• Wait a (short) random period before retransmitting if channel is idle (reduces
collision likelihood)
• Counter-based scheme – if a node hears more than k-neighbours broadcasting a
given RREQ before it has chance to forward, do not forward it yourself (the nodes
neighbours have probably seen the RREQ by now anyway)
• Distance based scheme – if a node within a given physical distance has already
transmitted, don’t forward – remember wireless networks are broadcast, every node within coverage range will hear, so if coverage ranges have significant overlap, no point retransmitting
Activity
Consider the network below, how would DSR determine a route from S to D? If the link between C and G were to break, how would that impact on the final route selection?
Y
B
S
L
CM J
Z
F
E
AHG
K
Route discovery:
• When node S wants to route data to node D, it initiates a route discovery
o S floods the network with RREQ messages
o Each node forwards the RREQ, appending its identifier
• Destination D on receiving first RREQ sends a route reply (RREP)
o RREP is sent on route obtained when RREQ path is reversed
o RREP includes the route from S to D on which RREQ was received
Data transmission:
• Node S, on receipt of RREP, caches the route
o Data for D is transmitted with entire route in header § Hence “source routing”
o Intermediate nodes remove their identifier from route and forward to next node
If the link between C and G were to break:
• C would issue a RERR message back towards source S
• Nodes hearing RERR remove C-G in their local caches
• A new route discovery takes place, likely going via H or B depending on link
charateristics
D
N
I