1. 1 MANETS – An overview

2. 2 MANETs Model and Definitions Simulatability – mobility models Routing algorithms Security issues with routing algorithms Tracing malicious faults Multipath & Multichannel routing

3. 3 Model and Definitions Definition 1 : Model of an ad hoc network Let V be a finite state system with state space S. The elements of V are mobile nodes: -- each node is a probabilistic finite state machine A mobile ad hoc network is a random process: G = { ( G 1, S 1 ), {( G 2, S 2 ), … … {( G T, S T ) } where the G 1 = ( V, N t ) are graphs with node set V and S t  S, subject to the following constraints:

4. 4 Model and Definitions Markov constraint – given ( G t, S t ) the next state ( G t+1, S t+1 ) is independent of all previous states ( G 1, S ! ) … … ( G t-1, S t-1 ) Mobility constraint –The transitional probabilities Pr [( G t+1, S t+1 ) | ( G t, S t )] are independent of time t. The distribution  generated by these is called the mobility distribution.

5. 5 Model and Definitions Medium constraint The communication medium is –is promiscuous –has limited bandwidth –is bidirectional

6. 6 Model and Definitions Definition 2 : Simulatability G is simulatable if there is an efficient algorithm  that simulates G, that is generates samples ( G* 1, S* 1 ), …, {( G* t, S* t ) such that Pr[( G* 1, S* 1 ), …, {( G* t, S* t )] = Pr[( G 1, S 1 ), …, {( G t, S t )]

7. 7 Model and Definitions Definition 3 : Route A route R ( s,d ) with source s and destination d is a list of nodes of G that start at s and end at d through which packets are forwarded. The nodes need not know the list, not even their successor on the list!

8. 8 Model and Definitions Definition 3 : Adversary Let  be a family of subsets V’ of V. We call  an Adversary Structure. The adversary Adv = Adv  selects a subset V’ of V and can corrupt all its nodes during the lifetime of the system. These nodes are called c orrupted or faulty. Adv is called a  - adversary..

9. 9 Model and Definitions Definition 3 : Adversary -- continued The adversary Adv may be passive or active. A passive adversary eavesdrops on the network communication. An active adversary uses the corrupted nodes to prevent the normal functioning of the network.

10. 10 Model and Definitions Definition 3 : Adversary -- continued The Byzantine threats model For this model  = { V’  V | | V’ | ≤ k } for some threshold k. In this case the adversary can corrupt up to k nodes.

11. 11 Mobility models The Random Walk model Nodes move in turns Each node at the beginning of its turn selects at random: –a value v for its velocity –a value  for its direction, and –a value t for the duration of the turn. –Alternatively instead of t a value d for the distance to be traveled is selected. Problem: Brownian motion

12. 12 Mobility models The Random Waypoint model An extension of the Random Walk model. Each node at the beginning of its turn first moves to a new position selected at random in the unit square. Problem: This addresses the Brownian motion, but now Nodes congregate to the centre of the unit square. Solution: Wrap around the parallel edges of the square to get a torus: now nodes can cross over boundaries.

13. 13 Mobility models The Random Waypoint model Problem: There is an speed decay for the nodes. Solution: Use special distribution of nodes at start time, minimum speed and warm up period. Problem: Abrupt changes in node direction at the beginning of turns.

14. 14 Mobility models The Random Direction model An extension of the Random Waypoint model. Each node at the beginning of its turn first –chooses -- see the Notes for a more control between the deterministic and random movement of nodes

15. 15 Mobility models Group models Group behavior is simulated by using a common reference point. As the reference point moves, the nodes move along with it.

16. 16 Mobility models Nomadic group model Nodes move randomly around the common reference point.

17. 17 Mobility models Target group model The common reference point is used as a target.

18. 18 Mobility models Hybrid models The movement of the common reference point is simulated by using one of the first three non group models. All these models are Markovian

19. 19 Mobility models Ad Hoc Network simulators Discussion on (and toy implementation?) of: –ns2 –OPNETR modeler –GLoMoSim

20. 20 Routing algorithms Definition Routing = ( Pathfinder, Communication ) Pathfinder is a distributive algorithm that takes as input G, s,d and generates a path  that links s,d. Path i nformation  xi is then distributed to all nodes on the path –With source-centric* routing, the source gets the complete path (from neighbor lists). –With network-centric* routing the path information is distributed among the nodes of  (as distance vectors). ----------------------------------- See later for definitions

21. 21 Routing algorithms Communication is a distributed algorithm on . On input data to s : – s broadcasts pkt [ s ] = ( header [ s ], data ), –If x i  s,d on  receives pkt [ x j ], it verifies it, and if correct computes and broadcasts: pkt [ x j ] = ( header [ x j ], data ). –If d receives pkt [ x k ], it verifies it, and if correct outputs data. header [ x ] contains information about the path  known to x. From this information, and from the local path information distributed by Pathfinder, any neighbor y of x can determine whether it is on , and if so, it computes header [y] and broadcasts pkt [ y ]. Eventually, if s,d are linked in G, data will be delivered to the destination d.

22. 22 Routing algorithms Definition Routing is reliable if: –when Adv is passive, on input G, s,d : any data input to s will be delivered to d. Routing will tolerate , if it is reliable in the presence of any active  - adversary.

23. 23 Routing algorithms Definition Routing updates Route settling

24. 24 Routing algorithms Definition There are two types of routing: network-centric and source-centric. With network-centric routing the routing effort is distributed within the network –discussion on tables and vectors. With source-centric routing most of the effort is done by the source, with the other nodes restricted to relaying packets and usually making only minor checks.

25. 25 Routing algorithms Discussion on basic source-centric routing algorithms: –DSDV, AODV, WRP, etc (Tina Suen) Discussion on basic network-centric routing algorithms –DSR etc

26. 26 Security issues for routing algorithms From a security point of view network-centric routing requires substantial cooperation between nodes and strong trust relations. These algorithms are therefore more vulnerable to malicious faults Source centric routing lessens the dependency on node cooperation and therefore is less vulnerable to malicious attacks.

27. 27 Security associations and PKIs Crypto mechanisms are used to support the basic security requirements for ad hoc networks. Whether secret key or public key systems are used security associations must be established between the nodes of the network. One way is to use an external PKI Other ways to be discussed in class.