ITIS 6010/8010
Wireless Network Security
Dr. Weichao Wang
• AODV (Ad hoc On-demand Distance Vector)
– On-demand protocol: routes are established when
needed, nodes not on active paths do not have to
maintain any information, routes will expire if not used
– Using hello messages to discover local topology (why
DSDV does not need this)
– Routes have lifetime
– Using sequence numbers to prevent loop
– Still a table driven protocol
2
• Three kinds of packets:
– Route request (RREQ)
– Route reply (RREP)
– Route error (RERR)
• Path discovery
– Every node has two value: sequence number
and broadcast id
– Broadcast RREQ: <s, d, seq_s, seq_d,
broad_s, hop>
3
• The pair <s, broad_s> uniquely identifies the
RREQ. Broad_s will be incremented every time
a RREQ is sent.
• Processing of RREQ
– Remembers reverse path to s: s, seq_s, hop count,
previous node, route lifetime
– If it has an active route, will reply with RREP
– Otherwise, rebroadcast RREQ and increase hop by 1
• Expanding ring search for destination
4
• Sending a RREP
– The RREP must have a fresher sequence
number of destination
– The RREP is unicast back to the source: <s,
d, seq_d, hop count, lifetime>
– How can the intermediate nodes figure out
hops to destination?
– The nodes along the path can setup forward
routes to the destination
– The fresher route is preferred over short route
5
• Routing entry contains
– Destination, next hop, hop count, seq_d, lifetime of
the path, neighbors have used this route
• Link failure response
– RERR is sent back with incremented sequence
number and infinity hop count
– All active routes using this next hop will get a copy of
this RERR and the information will be propagated
– New RREQ or local repair can be adopted
• Does AODV support multiple paths b/w a source
and a destination?
6
7
• DSR (Dynamic Source Routing)
– An on-demand routing protocol
– Nodes gather topology information by
overhearing network traffic
– Achieves much less control traffic compared
to DSDV (attention here)
– Source routing is used in every data packet
– Achieve loop-free
– Support unidirectional routes
8
• Route discovery
– A node may discover and cache multiple routes to a
destination
– Route request packet: <s, d, broad_s>
– The neighbor will
• If knows a route to d: send back a reply with the full route
• Otherwise, add its own node id and broadcast again
– The node IDs show the accumulated path from the
source and can be used to send route reply
9
• Route cache by overhearing
– May get multiple routes to the same destination
– Accelerate route discovery
– Cut unnecessary intermediate nodes
• Ensuring packet delivery
– Active acknowledgement
– Passive acknowledgement
• Preventing route reply storms
• Route error discovery and propagation
– Attach the broken link in new RREQ to prevent it from
being reused
10
• Advantages and disadvantages
– Simple routing protocol with low overhead
– High latency in finding routes
– Not scalable
11
12
• Zone Routing Protocol (ZRP)
– Reactive and proactive protocol each has
advantages and disadvantages
– Can we combine them?
– For nearby nodes, use proactive approach;
for faraway nodes, use reactive approach
– ZRP has a flat view of the network. It is not a
hierarchical protocol
13
• Concept of zone
– Every node has its own zone with a radius of r
hops
– Zones of different nodes will overlap
– Peripheral nodes and interior nodes
• Routing:
– Within the zone, use proactive (intrazone)
– Outside of the zone, use reactive (interzone)
– Each can be a family of routing protocols
14
• Discovery of local topology
– Hello beacons to detect active neighbors
– Zone notification messages to determine
nodes in zone
• Intrazone routing
15
• Interzone routing
– Instead of broadcast, the RREQ is sent to
peripheral nodes through bordercast
• Can be achieved through multicast or multiple
unicast
• If the peripheral node knows the route, sends back
a reply (source routing or previous hop of the
peripheral nodes)
• Otherwise, bordercast again
– Sharply reduce overhead compared to
broadcast
16
• The size of a zone
– If r = 1, it goes back to pure reactive
– If r = ∞, it goes back to proactive
• Route maintenance
– When a link breaks, the node can find an alternative
route based on its zone knowledge
• Overhead control
– Since zones overlap, a node may receive the same
RREQ multiple times
– How to guarantee the RREQ only propagate to outer
areas?
17
• Overhead control
– Query detection
• If a node has forwarded or overheard a RREQ, it
does not transfer in the reverse direction
– Early termination
• If a node knows that the target of bordercast
already gets the RREQ, it discards the packet
– Random delay of bordercast
18