1. CSE 4340/5349 Mobile Systems Engineering
M. Kumar
Spring 2010
Week 7 Mobile Ad Hoc Networks (MANETs)

2. Organization Mobile Ad hoc Networks Routing Service Discovery
Trust and Authentication

3. What are Ad Hoc Networks
Infrastructure-less mobile networking
A collection of mobile nodes forming a temporary network without a fixed infrastructure
Examples
In airplanes, ships, trucks, cars
On people, personal area networks (PANs)
Sensor networks
Nodes move around arbitrarily
A MANET system may operate in Isolation
MANET nodes are equipped with wireless transmitters and antennas
Energy-constrained operation

4. Topology Dynamic Decentralized Self-organizing Multihop
Mobility, disconnection
Decentralized
Peer-to-peer connectivity
Self-organizing
Adapt to changing situations
Multihop
Intermediate nodes receive and forward messages

5. Main Challenges Routing Security Resource Limitations
No infrastructure, constantly changing nodes
Mobility is viewed as a challenge
Every a node joins or disconnects
House keeping jobs
Security
New vulnerabilities, nasty neighbors
Resource Limitations
Batteries, limited computing power

6. Routing No infrastructure Source to destination route
Participating nodes must perform routing functionalities.
Source to destination route
Comprises many nodes
Receive and forward packets
Many potential routes
the best path must be chosen

7. Routing Protocols Distributed operation Loop-freedom
Packets spinning around in the network
TTL may be one solution
Demand-based operation
Adapt to the traffic patterns based on demand
Intermediate nodes
energy constrained
mobile

8. Ad Hoc Routing Protocols
Table Driven
Source-Initiated
DSDV
WRP
AODV
DSR
LMR
ABR
CGSR

9. Table driven protocols
Each node to maintain one or more tables for storing up-to-date routing information.
Respond to changes in the network topology by propagating updates throughout the network.
Various protocols
Differ in the number of routing related tables required and the methods of network change broadcasting.

10. Source-initiated or on-demand routing
Creates route only when desired by a source node.
The source node initiates a route discovery process.
A route is found after examining all possible route permutations.
This route is maintained for the duration of the current routing

11. Destination Sequenced Distance Vector Routing (DSDV)
Hop-by-hop distance vector routing protocol
Loop freedom guaranteed
Sequence number used for making decisions
Periodic route updates
Complexity grows as O(n2)

12. Clusterhead Gateway Switch Routing (CGSR)
CGSR is a clustered multihop network
DSDV used as underlying routing protocol
A cluster head table is necessary in addition to the routing table
Hierarchical cluster-to-gateway routing approach to route traffic from source to destination

13. Clusterhead Gateway Switch Routing (CGSR)

14. Dynamic Source Routing (DSR)
Intended for networks in which the nodes move at moderate speed
No need for intermediate nodes to maintain up-to-date routing information
Involves route discovery and route maintenance

15. On Demand Multicast routing protocol (ODMRP)
On-demand technique to establish membership
Use broadcast nature of links
Reactive beacons
Sender has to join multicast group
Creates a mesh of nodes to provide redundant multicast routes
Routing decision based on flag settings

16. Ad-Hoc On-Demand Distance Vector (AODV)
Combination of both DSR and DSDV
Assumes symmetric links between the nodes
Route discovery and route maintenance from DSR
Hop-by-hop routing, sequence numbers and periodic updates from DSDV
Support multicasting

17. Comparison of routing protocols
On demand
Routing information
Periodic routing updates are not needed
Traffic overheads are created when needed
Slow to adapt
Table driven
Always available regardless of need
Periodic routing updates are needed
Traffic overheads when nodes move
Regardless of need
Quick to adapt

18. Address Auto-configuration
Stateless approaches
No allocation states
high overhead, scalability problems
Stateful approaches
Central allocation states
Distributed global allocation states: High communication overhead/delay, MANETConf
Local allocation states: limitations in dynamic situations,
Stateless approaches
No allocation states: new joining nodes self-configure themselves
DAD (Duplicate Address Detection) incurs high overhead leading scalability problems
Auto-configuration for IPv4, a stateless auto-configuration mechanism for IPv6, IP address auto-configuration for ad-hoc networks
Stateful approaches
Central allocation states
Distributed global allocation states: High communication overhead/delay, MANETConf [6]
Local allocation states: limitations in dynamic situations, a buddy system [7]

19. Service discovery Two Approaches
Centralized systems
Decentralized systems
Both systems are unsuitable for pervasive environments
Static servers and administration are not expected semi- permanently
Not scalable due to high overhead of multicast or broadcast
Standards: Sun’s Jini, SLP of IETF, Microsoft’s UPnP
Service Discovery in Ad hoc networks:
lack flexibility or highly complex
Others: Intentional Naming System (INS), INS/Twine, Secure Service Discovery Service (SDS)
Two Approaches
Centralized systems [20, 21]
Decentralized systems [15, 21, 22]
Both systems are unsuitable for pervasive environments
Static servers and administration are not expected semi-permanently
Not scalable due to high overhead of multicast or broadcast
Standards: Sun’s Jini, Service Location Protocol (SLP) of IETF, Microsoft’s Universal Plug and Play (UPnP)
Service Discovery in Ad hoc networks: lack of flexibility or employs high complexity
Others: Intentional Naming System (INS), INS/Twine, Secure Service Discovery Service (SDS)

20. Trust management and Authentication
Centralized and Distributed approaches
Distributed trust management systems
Each node performs trust management locally
High overheads and complexities associated with each individual entity
Authentication in open environments
Based on physical inspection or social relationships
Use asymmetric cryptosystems due to difficulty of key management
Authentication and trust management are dependent
Two approaches: Centralized and Distributed
Distributed trust management systems [50, 68, 72, 74, 85, 86, 95]
Each node performs trust management locally
High overheads and complexities associated with each individual entity
Authentication in open environments is usually achieved in a distributed manner [57, 58, 78, 93, 94]
Based on physical inspection or social relationships
Use asymmetric cryptosystems due to difficulty of key management
Authentication and trust management are dependent
SECURE project [68, 92] proposes entity recognition [73] as a general substitution of authentication but not scalable due to high complexities

21. Applications of MANETs
Personal Area Networks
Vehicles
Crisis Management
Sensor Networks