Analysis of a SuperSEAD
Aaron Staple
Mukund Sundararajan
Mobile Ad-hoc Networks
Changing topology
Relatively Low Power :No Asymmetric Crypto
Low Physical Security
Broadcast physical medium
Group of people with laptops in a room
Routing Protocol: DSDV
Distance Vector Routing = Distributed Bellman-
Ford
Sequence Numbers prevent Routing Loops
Routing Table: Contains ID,Metric, SequenceNo,
NextHop
Periodic Updates: Sequence Numbers
Higher Sequence Numbers, Lower Metrics take
precedence
Contextual assumptions
Wireless Links are Bidirectional
Physical Layer attacks are beyond the scope of
the Protocol – Jamming, DOS
Number of Nodes is known and no new nodes
can be added to the network.
Routing information is distributed via broadcasts
Attacker Actions
Failing to Advertise Routes
Ignoring existing routes
Modifying routing updates :Black holes
Replaying information
Wormhole Attack
Assumptions regarding Attacker
Power
Attacker nodes have the same capabilities as other
nodes
Cannot schedule arbitrary inter-leavings
Can talk to nodes in its vicinitae
Cant disrupt other nodes messages
Compromised Nodes
Compromised Key material
Collusion
Dolev-Yao attackers.
What SuperSEAD attempts to
accomplish.
“SEAD performs well over the range of
scenarios we tested,and is robust
against multiple uncoordinated
attackers creating incorrect routing
state in any other node,even in spite of
any active attackers or compromised
nodes in the network.”
Secure Efficient Ad Hoc Distance Vector
Routing
SuperSEAD Protocol
Hash Tree Chains to Authenticate the lower bound on
the metric and an upper bound on the sequence
number
Neighbor Authentication:
Origin of the message
N^2 Symmetric keys
Hash Chain Anchors and Symmetric Keys are
distributed using an external mechanism
Packet Leashes :Temporal
Avoid replays.
Hash Chains Metric
S 0 1 2 3 4
e
q
u
e 3
n
c
e 2
N
O
1
Hash Trees
Hash Tree Chains
Hash Trees Chains Contd..
Analysis Method :Equivalence Based
Impact on network topology achieved in the
presence of a few attacker nodes
Compare states achievable in the presence of
attacker to an attacker-free model
We consider only steady states
Essentially simulating the hash chains and the
neighbor authentication, assuming that they
operate correctly
Security Properties
Correct Routing state at all good nodes about all the
other good nodes
Review of Attacker capabilities:
Cant interfere with any messages
Cant advertise different information to two different neighbors.
Cant perform arbitrary inter-leavings of messages
Cannot determine all the events that take place in the
network
Murphi Transition Scheduling
We DO NOT consider all possible schedulings of
routing updates
Attackers can’t control the scheduling
Murphi State Space would be extremely large
We randomly generate permutations that
determine order of broadcast updates
All attacker actions are enumerated
Project Flow
Generate sequence of
topologies
All-pairs Shortest Generate Murphi
Paths Code
Simulate/Verify
Model
Analyze resulting steady
states Generate
Message Trace
for Offending
states
ruleset change: boolean do
ruleset new_seq_no: 0..MaxSequenceNumber do
ruleset new_distance: 1..(MaxDistance-1) do
rule 300 "A bad node performs a broadcast update about a single other node"
(turn_list[turn] > NumGoodNodes)
==>
begin
for j: 1..TotalNodes do
if ((topology[top_id][turn_list[turn]][j] = true) &
((routing_tables[turn_list[turn]][badAbout].sequence_no > new_seq_no) |
((routing_tables[turn_list[turn]][badAbout].sequence_no = new_seq_no) &
(routing_tables[turn_list[turn]][badAbout].distance + 1 <= new_distance))) &
((routing_tables[j][badAbout].sequence_no < new_seq_no) |
((routing_tables[j][badAbout].sequence_no = new_seq_no) &
(routing_tables[j][badAbout].distance > new_distance)))) then
routing_tables[j][badAbout].sequence_no := new_seq_no;
routing_tables[j][badAbout].distance := new_distance;
printout := 1;
end; end;
if (badAbout = TotalNodes) then
turn := (turn % TotalNodes) + 1;
change_top := change;
end;
badAbout := (badAbout % TotalNodes) + 1;
end; end; end; end;
Attacks that we focused on
Run the protocol without SEAD present and see the attack
Run it with SEAD present and if an attacker node cannot
advertise different information to different neighbors, show that
found no attack.
In the presence of collusion,tunneling there is a wormhole attack.
Node placement attack.
K (>1) consecutive attacker nodes on a path can shorten path by
k-1.
Attacks in the absence of neighbor authentication and
packet leashes
Issues that we faced
Inconsistently specified Murphi Syntax [Some
things that could have been wrong were wrong
at the worst possible time]
Difficult to model a more representative subset
of all possible routing update inter-leavings.
No Protocol Specification, Only Prose
Had to modify poorly documented Murphi Code.
Conclusions
Tool Related
Simulate certain moves but verify other moves
Should scale to verifying larger collections of nodes
Connectivity is orthogonal to the protocol and it is useful to be able to
specify separately
Print out all states that satisfy certain conditions
SEAD Related
New nodes cannot join the network
Simple collusion attacks
Need for reputation mechanisms
Lot of assumptions at the physical layer
Attacker power
References
1. SEAD:secure efficient distance vector routing for mobile wireless ad
hoc networks
Yih-Chun Hu,David B.Johnson,Adrian Perrig
2. Y.-C.Hu,A.Perrig,D.B.Johnson,Packet leashes:a
defense against wormhole attacks in wireless ad hoc
networks,in:Proceedings of IEEE Infocomm 2003,April
2003.
3. Efficient Security Mechanisms for Routing Protocols
Yih-Chun Hu ,Adrian Perrig,David B. Johnson
4. Secure Routing in Wireless Sensor Networks:Attacks and Countermeasures
Chris Karlof David Wagner
5. The TESLA Broadcast Authentication Protocol
Adrian Perrig Ran Canetti J. D. Tygar Dawn Song
