View
214
Download
0
Tags:
Embed Size (px)
Citation preview
04/18/23 1
Mobile Ad-hoc Networks: Issues and Challenges
Uyen Trang NguyenDept. of Computer Science & EngineeringYork University (Toronto, Canada)
04/18/23 2
Wireless Networks: TaxonomyWireless Networks
Infrastructure-based Infrastructureless
Cellular Networks
Wireless LANs Ad Hoc NetworksMobile Adhoc Networks
Static NodesMobile Nodes
Sensor Networks
04/18/23 3
MANETs: Introduction No infrastructure (no base stations or access
points) Mobile nodes
Form a network in an ad-hoc manner Act both as hosts and routers Communicate using single or multi-hop wireless links
Topology, locations, connectivity, transmission quality are variable.
04/18/23 5
Applications Civil
Disaster recovery Taxi cabs Communications over water using floats Vehicular ad-hoc network
Defense Battlefield communications Monitoring and planning
04/18/23 6
Challenges Unpredictable mobility Wireless channels: error-prone media Low bandwidth channels Devices: low power, limited resources Maintaining connectivity, states Security
04/18/23 7
Issues to Be Discussed Medium access control Transport layer issues Security Incentives for cooperation
04/18/23 8
MAC for MANETs: Requirements To avoid interference among simultaneous
transmissions But enable as many non-interfering transmission as
possible Maintain fairness among transmissions
No centralized coordinators: fully distributed operations
No clock synchronization: asynchronous operations
04/18/23 9
Carrier Sensing in MANETs Problems
Hidden terminal problem Exposed terminal problem
Sensing range Transmission range
Contention matters only at the receiver’s end
04/18/23 12
MACs Suitable for MANETs MACA [Karn 1990]
Proposes to solve the hidden terminal problem by RTS/CTS dialog
MACAW [Bharghanvan 1994] Increasing reliability by RTS/CTS/DATA/ACK dialog
IEEE 802.11 Distributed Coordination Function (DCF) Also use RTS/CTS/DATA/ACK dialog
04/18/23 14
RTS/CTS dialog (2)
RTS
Defer
CTS
Defer
Any node hearing this CTS will defer medium access
04/18/23 16
IEEE 802.11 DCF Uses RTS/CTS exchange to avoid hidden
terminal problem Any node overhearing a CTS cannot transmit for the
duration of the transfer. Any node overhearing an RTS cannot transmit for the
duration of the transfer (to avoid collision with ACK) Uses ACK to achieve reliability CSMA/CA
Contention-based random access Collision detection not possible while transmitting
04/18/23 17
IEEE 802.11 DCF (cont.) Carrier sense in 802.11
Physical carrier sense Virtual carrier sense using Network Allocation Vector
(NAV) RTS/CTS specify duration of subsequent DATA/ACK NAV is updated based on overheard RTS/CTS
Collision avoidance Nodes stay silent when carrier sensed busy
(physical/virtual) Backoff intervals are used to reduce collision
probability
04/18/23 18
Backoff Interval When channel is busy, choose a backoff
interval in the range [0, cw]. Count down the backoff interval when medium
becomes idle. Count down is suspended if medium becomes
busy again. When backoff interval reaches 0, transmit RTS. Binary exponential backoff in 802.11 DCF:
When a node fails to receive CTS, cw is doubled up (up to an upper bound).
When a data transfer completes successfully, cw is reset to cwmin.
04/18/23 20
Disadvantages of IEEE 802.11 DCF
High power consumption Hidden terminal problem not totally solved
(e.g., collision of RTS) Exposed terminal problem not solved Fairness problem among different transmitting
nodes Only providing best-effort service
04/18/23 21
MAC for Multicast: a Challenging Issue Multicast: efficient info
delivery from a source to a set of destinations simultaneously
Uses 802.11 CSMA/CA
Cannot use RTS/CTS exchange
Currently there are no effective MAC protocols for multicast
04/18/23 22
Issues to Be Discussed Medium access control Transport layer issues Security Incentives for cooperation
04/18/23 23
TCP in Wired Networks Receiver sends ACKs for packets received correctly
Sender times out on unacknowledged packets, retransmits Sender adjusts congestion window
04/18/23 24
TCP in MANETs
TCP is designed for wired networks
Low bit error rate Loss mainly caused
by congestion Routes relatively
fixed
TCP in MANETs
High bit error rate Unreliable wireless
channels Route changes due
to node mobility
04/18/23 25
Consequences TCP sender misinterprets losses as congestion
retransmits unacknowledged segments Why retransmit when there is no route
invokes congestion control enters slow start recovery
Throughput is always low
Why use TCP at all in such cases? For interactions with the Internet and seamless
portability to applications using standard TCP (file transfer, email, browsers).
04/18/23 26
Approaches to Improving TCP Hide error losses from the sender
Sender will not reduce congestion window
Determine the cause of loss If due to errors, do not reduce window size
Modifications are done at the sender only the receiver intermediate nodes only combinations of the above
04/18/23 28
ATCP Approach Uses network layer feedback from
intermediates nodes for appropriate actions Different types of network feedback:
ICMP: “Destination Unreachable” message indicates route break Stops transmission Waits until a new route is found and resumes trx
ECN: ACK with ECN flag indicates network congestion Invokes congestion control
Retransmission time-out or 3 duplicate ACKs: Retransmits unacknowledged segments without
shrinking congestion window
04/18/23 29
Multicast Transport Reliable delivery
Feedback: ACK, NACK Congestion control
Feedback: loss rate, data rate Group-based
not scalable feedback implosion exposure
Tree-based scalable feedback implosion solved limited exposure
04/18/23 30
Multicast Transport in MANETs: a Challenging Issue
Tree-based extremely difficult to establish a tree-based structure
due to node mobility Group-based
not scalable feedback suppression: random timer, probabilistic multiple simultaneous transmissions: potential
collisions Calculation of round-trip-time: difficult due to
node mobility route break
04/18/23 31
Issues to Be Discussed Medium access control Transport layer issues Security Incentives for cooperation
04/18/23 32
Vulnerabilities of MANETs Wireless links jamming Broadcast nature eavesdropping Mobility, dynamics difficult to detect
anomalies (e.g., bogus routes) No central authorities or infrastructures
difficult key management Trade-offs between resource constraints and
security
04/18/23 33
Potential Attacks Impersonation
An attacker assumes identity and privileges of an authorized node
Denial of service (e.g., jamming, flooding) Network layer attacks
Blackhole a node falsely advertises good paths drops packets
Wormhole: 2 colluding attackers form a tunnel Byzantine: creating routing loops, non-optimal paths Resource consumption attack: generating unnecessary
packets (false route requests, beacons)
04/18/23 34
Attack Countermeasures Defense against external attacks
Authentication + encryption (needs key and trust management)
Defense against internal attacks Secure routing protocols (e.g. SEAD, ARAN, SAODV) Protect routing metrics by hashing
Intrusion detection techniques CONFIDANT
04/18/23 35
Security: Open Issues Key management
Still relies on a central authority (or a group of trusted nodes)
Consumes high power for processing Intrusion detection
Can only detect a group of potential attackers Cannot single out the true attackers or compromised
nodes High rate of false alarm due to unreliable wireless
links and randomness of channel access No solutions to flooding/jamming attacks Incentives for nodes to cooperate
04/18/23 36
Issues to Be Discussed Medium access control Transport layer issues Security Incentives for cooperation
04/18/23 37
Incentives Rewards: per-hop payment in every packet or
in counters embedded in nodes may not always be effective
Punishment punishing both selfish and malicious nodes Example: CONFIDANT [Buchegger 2002]
Based on the biological example in “The Selfish Gene” by Richard Dawkins (Oxford University Press, 1989 edition, 1976)
04/18/23 38
“The Selfish Gene” Reciprocal altruism is beneficial for every
biological system when favors are granted simultaneously
Example: survival chances of birds grooming parasites off each other’s head.
3 types of birds: Suckers: always help others Cheats: get help from but never help others Grudgers: start out by helping every bird
but bears a grudge against those that do not help and subsequently no longer groom their heads
04/18/23 39
CONFIDANT
Every node has 4 components: The monitor The trust manager The reputation system (node rating) The path manager
04/18/23 40
CONFIDANT Components The monitor: neighborhood watch
Listening to transmissions of next nodes Observing route protocol behaviors (no forwarding,
unusual route attraction, unusually frequent route updates)
Calling the reputation system when a bad behavior is detected
The trust manager: distributed and adaptive Using trust function to calculate trust levels Forwarding ALARM messages to warn others of
malicious/selfish nodes Filtering ALARM messages based on trust level of the
reporting node
04/18/23 41
CONFIDANT Components (cont.)
The reputation system (node rating): based on Own experience: greatest weight Observations: smaller weight Reported experience (from ALARM messages): weight
function according to trust level Rating lists are maintained locally, may be
exchanged with friends The path manager
Deleting paths containing malicious nodes Dealing with route requests from malicious nodes
(e.g., ignore, alert the source)
04/18/23 42
Disadvantages of CONFIDANT Authentication is a prerequisite
Requires efficient key management
Implementing all the components requires high processing power and storage incurs lots of overheads
04/18/23 43
SummaryMANETs Unpredictable mobility Low bandwidth channels Wireless channels: error-prone media, vulnerable to
attacks Devices: low power, limited resourcesIssues Medium access control Transport layer issues Security Incentives for cooperation