1 Ad Hoc Networks Security Instructor: Carlos Pomalaza-Ráez Fall 2003 University of Oulu, Finland.

1

Ad Hoc Networks Security

Instructor: Carlos Pomalaza-Ráez

Fall 2003University of Oulu, Finland

2

Introduction to Cryptography The idea is to protect data by transforming into a representation from which is hard to recover. This provide us with: Confidentiality – only the sender and the receiver should know the

message content

Authentication – sender and receiver can confirm the identity of each other

Integrity – sender and receiver can detect any alteration of the message

Non-repudiation – sender can not deny having created the message

Freshness – message is recent and not a replay

Unless a message is properly protected unfriendly “agents” can capture or see it as it moves across the network and, Insert messages into the connection

Impersonate – fake (spoof) source address

Hijack – take over connection a replacing the sender or receiver

Denial of service – by, for example, overloading the resources

3

Private (Symmetric) – Key systems

In these systems the message M is encrypted using a key e which is known only to the sender and the receiver. To encrypt the message compute X = E(M, e), E being the encryption function. To decrypt X compute M = D(X, d), where d is the decryption key corresponding to e. There is usually a simple relationship between e and d. A widely known secret-key system is DES (Data Encryption Standard)

M encryptionalgorithm

decryption algorithm

e – encryption key d – decryption key

M = D(X, d)

Alice

X = E(M, e)

Bob

Unfriendly agent Eve

4

Public (Asymmetric)– Key systems

In these systems the message M is encrypted using a key e which is public. To encrypt the message compute X = E(M, e), E being the encryption function. To decrypt X compute M = D(X, d), where d is the decryption key corresponding to e. Knowing e doesn’t help anyone to discover the decryption key d.

M encryptionalgorithm

decryptionalgorithm

e – public encryption key d – private decryption key

M = D(X, d)

Alice

X = E(M, e)

Bob

Unfriendly agent Eve

5

RSA – A Public-key Crypto-System

RSA stands for its inventors Ron Rivest, Adi Shamir, and Len Ademan. We assume here that message is broken into parts of the right size, e.g. 1024 bits.

Choosing Keys

Choose two large prime numbers p, q (e.g., 512 bits each)

Compute n = pq, z = (p-1)(q-1) = Ф(n)

Choose e, (e<n), such that it has no common factors with z, e.g. gcd(e,Ф(n)) = 1 (gcd stands for greatest common divisor)

Choose d such that ed-1 is exactly divisible by z (in other words,ed mod Ф(n) = 1)

Public key is (n,e)

Private key is (n,d)

R.L. Rivest, A. Shamir, L. Adleman, “A Method for Obtaining Digital Signatures and Public-Key Cryptosystems,” Comm. of ACM, 21 (2), pp. 120-126, Feb. 1978.

6

RSA: Encryption - Decryption

)(mod

computes,sender theRSA, using message a send To

nMX

Me

MnMnM

nednM

nMnXM

ed

n

edd

)(mod)(mod

),( of multiple some 1 is and 1)(mod Since

)(mod)(mod

computes,receiver themessage edecrypt th To

1

)(

0

)(mod)),(,()),(,(

:is that identical, are functions decryption and encryption that Notice

nXnkXDnkXE k

7

Hash AlgorithmsA basic tool for cryptography is a secure hash algorithm. Given a variable length message x, a secure hash algorithm computes a function h(x) which has a fixed and often smaller number of bits. It is usually not possible to recover x from its hash function. Desirable properties of a secure hash function are:

A hash function h(x) is one-way if given y it is hard to find x such that h(x) = y

A hash function h(x) is weakly collision-free if given a message x1 it is hard to find another message x2 such that h(x1) = h(x2)

A hash function h(x) is strong collision-free if it is hard to find any pair of messages x1, x2 such that h(x1) = h(x2)

An important property of secure hash functions, like any hash function, is that they should uniformly cover their range. That is, for a uniform distribution of the inputs, the output probabilities from the hash function should be uniform.

8

Authentication via Digital Signatures

Method I: Bob encrypts entire message with his private key; this is Bob’s digital signature

Bob send both the message and his digital signature

Similar to handwritten signatures

Dear Alice,

Bob Bob’s private key

Ready for Transmission Alice decrypts Bob’s message using Bob’s public key

If decrypted message matches the message, Alice knows that the signed message could only have come from Bob

Signing the entire document/message is computationally expensive

Original Text

Text encrypted with Bob’s private key

9

Authentication via Digital Signatures

Method II:

Compute a hash on the document/message

The hash, also called a message digest, is much smaller than the document, resembles a CRC (Cyclic Redundancy Check)

Use private key to encrypt only the message digest

Encrypted digest is commonly called a digital signature

Computationally inexpensive

Send both the document and the digitally signed message digest

At receiver

Hash the document → MDA and decrypt the digital signature → MDB

If MDA = MDB then receiver knows that:

the identity of sender correctly matches the advertiser of the public key (authentication)

that the document hasn’t been tampered with (data integrity)

10

Digital Signature - Signed message digest

Bob sends digitally signed message

OriginalLong

MessageFixed-sizemessage

digest

Encryptedmessage

digest

Bob’sprivate

key

Send to Alice

Alice verifies signature and integrity of digitally signed message

Many to onehash function

Fixed-sizemessage

digest

OriginalLong

Message

Many to onehash function

Encryptedmessage

digest

Bob’spublic

key

Fixed-sizemessage

digest

Compare

11

Non-Repudiation via Digital Signatures

Fixed-sizemessage

digest

OriginalLong

Message

Many to onehash function

Encryptedmessage

digest

Bob’spublic

key

Fixed-sizemessage

digest

Compare

MDAMDB

Digital Signatures provide authentication, integrity, and non-repudiation

At receiver, if MDA = MDB then receiver knows that:

Only the sender’s private key could have created this signature (Non-repudiation & Authentication)

Sender can’t deny sending message

12

One-Way Hash Chains

Construction

Pick random rN and a public one-way function F

ri = F(ri+1)

Secret value: rN

Public value: r0

Fr8r5 r6 r7

F F Fr4

Properties

Use in reverse order of construction, i.e. r1, r2,…, rN

It is not feasible to derive ri from rj (j<i)

Good for authenticating ri knowing rj (j<i), e.g. verify if rj = Fi-j(ri)

Robust to missing values

13

Message Authentication Codes (MAC)

It is a code – MAC(K,M)

Calculated by some function MAC that requires little computation

Inputs are the message M to be sent and K, the symmetric key known only by the two parties

The code is appended to each packet, i.e. {M, MAC(K,M)}

BobAlice

ComputeC = MAC(K,M)

{ M | C }

It’s Alice not AliceYes No

ComputeC’ = MAC(K,M)

C = C’Alice and Bob know in advance K and the MAC function

14

Unicast Source Authentication

Bob

Dave

Carol

Alice

Ka-b

{M|MAC(Ka-b, M)}

Ka-b

Ka-cKa-d

M is duplicated and sent separately to each intended receiver with it a different MAC

High overhead and consumes network resources

Internet

{M|MAC(Ka-c, M)}

{M|MAC(Ka-d, M)}

15

Multicast Source Authentication

InternetInternet

Bob

Dave

Carol

Alice

Ka

{M|MAC(Ka, M)}

Ka

KaKa

Ka is known to all receivers. Any receiver can forge a packet

Low overhead and less network resources when compared with unicast method

16

TESLATimed Efficient Stream Loss-Tolerant Authentication Uses symmetric key cryptography Asymmetric key cryptography via time Based on initial loose time synchronization MAC is attached to each packet Delayed-disclosure of keys

MAC(Ki,M)

M

timeti-1 ti ti+1

F(Ki)Authentic

Commitment

Ki

is disclosed

1- Verify Ki

2- Verify MAC3- M is authentic

A. Perrig, R. Canetti, J.D. Tygar, D. Song, “Efficient authentication and signing of multicast streams over lossy channels,” IEEE Symposium on Security and Privacy, May 2000.

17

TESLA – Sender SetupAlice

time

interval i -1 interval i interval i +1 interval N

Ki+1KiKi-1 KN

Use F' to derive the key to compute MAC K‘i= F’(Ki)

K’i+1K’iK’i-1 K’N

F’ F’ F’ F’

Break time in intervals of same duration

Determine key chain length N, picks the last key KN randomly

Using a One Way Pseudo Random Function F compute Ki = F(Ki+1), assign one key to each interval

Key generation

18

TESLA – Authentication

Ki+1KiKi-1

K’i+1K’iK’i-1

F’ F’ F’

Mi-1, Ki-2 MAC(K’i-1, Di-1)

Di-1

Mi , Ki-1 MAC(K’i, Di)

Di

Mi+1, Ki MAC(K’i+1, Di+1)

Di+1

Pi-1 Pi Pi+1

authenticated authenticated after reception of Pi+1

not yet authenticated

When the receiver gets packet Pi,it can not verify the MAC since it does not yet know Ki from which it can compute K’i

Packet Pi+1 discloses Ki and allows the receiver to:

verify that Ki is correct, e.g., F(Ki) = Ki-1

compute K’i and check the authenticity of packet Pi by verifying the MAC of Pi

19

TESLA – Dynamic Packet Rates

Dj

K’i

Dj+1

K’i+2

i i +5i +4i +3i +2i +1

Dj+3

K’i+3

Dj+4

K’i+3

Dj+4

K’i+5

Mj+4

Ki+1

Mj+3

Ki-1

Mj+2

Ki-1

Mj

Ki-4

Mj+1

Ki-2

TΔ

d=4

The MAC key and the disclosed key depend of the time interval

The authentication key of Pj is Ki which is disclosed by packets sent in interval (i + d )

In this example packet Pj+4 discloses key Ki+1 which allows the receiver to compute Ki and to authenticate packet Pj

Pj Pj+2Pj+1 Pj+3 Pj+4

20

Attacks to Ad-Hoc Networks

Passive Only eavesdrop

Threats against privacy/anonymity

Active Injects packets and eavesdrops

Characterized based on the number of controlled nodes in the network

Routing disruption attacks Causes legitimate data packets to be routed dysfunctionally (e.g.,

routing loop, black hole, gray hole, detour, partition)

Resource consumption attacks Consumes valuable network resources or node resources (e.g.,

injecting data packets, injecting control packets)

21

ARIADNE: A Secure On-Demand Routing Protocol for Ad Hoc

Networks Withstands node compromise

Relies on highly efficient symmetric cryptography

Does not require trusted hardware or powerful processors

Authenticate routing messages using one of: Shared secrets between each pair of nodes

Avoids need for synchronization

Shared secrets between communicating nodes combined with broadcast authentication

Requires loose time synchronization

Digital signatures

Y-C Hu, A. Perrig, D. B. Jonson, “Ariadne: A Secure On-Demand Routing Protocol for Ad Hoc Networks,” 8th ACM International Conference on Mobile Computing and Networking, MobiCom 2002, pp. 12-23.

22

ARIADNE: Key Setup Shared secret keys

Key distribution center

Bootstrapping from a Public Key Infrastructure (PKI)

Pre-loading at initialization

Initial TESLA keys

Embed at initialization

Assume PKI and embed Certifications Authority’s public key at each node

Digital signatures

A mechanism distribute one authentic public key for each node

Each node also has an authentic element from the Route Discovery chain of every node initiating Route Discoveries

23

ARIADNE: Route Discovery Takes after DSR

Assume sender and receiver share secret (non-TESLA) keys for message authentication

Target authenticates ROUTE REQUESTS

Sender includes a MAC computed with end-to-end key

Target verifies authenticity and freshness of request using shared key

Data authentication using TESLA keys

Each hop authenticates new information in the REQUEST

Target buffers REPLY until intermediate nodes release TESLA keyso TESLA security condition is verified at the targeto Target includes a MAC in the REPLY to certify the condition was met

Attacker can remove a node from node list in a REQUEST, but

One-way hash functions verify that no hop was omitted (per-hop hashing)

24

ARIADNE: Route Discovery Assume all nodes know an authentic key of the TESLA one-way key chain of

every other node

Securing ROUTE REQUEST

Target can authenticate the sender (using their additional shared key)

Initiator can authenticate each path entry in the ROUTE REPLY

No intermediate node can remove any other node in the REQUEST or REPLY

ROUTE REQUEST packet contains eight fields: ROUTE REQUEST: label

initiator: address of the sender

target: address of the recipient

id: unique identifier

time interval: TESLA time interval of the pessimistic arrival time

hash chain: sequence of MAC hashes

node list: sequence of nodes on the path

MAC list: MACs of the message using TESLA keys

25

ARIADNE: Route Discovery Upon receiving ROUTE REQUEST, a node:

1. Processes the request only if it is new

2. Processes the request only if the time interval is valid (not too far in the future, but not for an already disclosed TESLA key)

3. Modifies the request and rebroadcasts it– Appends its address to the node list, replaces the hash chain with H[A, hash

chain], appends MAC of entire REQUEST to MAC list using KAi where i is the index for the time interval specified in the REQUEST

When the target receives the route request:

1. Checks the validity of the REQUEST (determining that the keys from the time interval have not been disclosed yet and that hash chain is correct)

2. Returns ROUTE REPLY containing eight fields– ROUTE REPLY, target, initiator, time interval, node list, MAC list– target MAC: MAC computed over above fields with key shared between

target and initiator– key list: disclosable MAC keys of nodes along the path

26

ARIADNE: Route Discovery Node forwarding ROUTE REPLY

Waits until it can disclose TESLA key from specified intervalo Appends that key to the key listo This waiting does delay the return of the ROUTE REPLY but does not

consume extra computational power

When initiator receives ROUTE REPLY

1. Verifies each key in the key list is valid

2. Verifies that the target MAC is valid

3. Verifies that each MAC in the MAC list is valid using the TESLA keys

The ROUTE MAINTENACE mechanism has similar features, e.g.,

Based on DSR Node forwarding a packet to the next hop returns a ROUTE ERROR to the

original sender

Prevent unauthorized nodes from sending errors, e.g. sender we authenticates errors