1. Weichao Wang, Bharat Bhargava Youngjoo, Shin 2006.09.12
Key Distribution and Update for Secure Inter-group Multicast Communication
Weichao Wang, Bharat Bhargava
Youngjoo, Shin

2. Contents Introduction Assumptions Straight forward approach
New approach
Secure group communication
Key update during group changes
Discussions
Conclusions
Key Distribution and Update for Secure Inter-group Multicast Communication

3. Introduction
Secure multicast has become an important component of many applications in wireless networks
Two types of group communications
Intra-group communication
Inter-group communication
This paper proposes a mechanism of key distribution and update for secure group communication
Intra-group communication
Inter-group communication
Key Distribution and Update for Secure Inter-group Multicast Communication

4. Assumptions Network and communication model Threat model
The links among wireless nodes are bidirectional
Two neighboring nodes can always send packets to each other
A centralized group manager (GM) is in charge of key distribution and key update
Threat model
Eavesdropping
Impersonation
Backward secrecy
Forward secrecy
Key Distribution and Update for Secure Inter-group Multicast Communication

5. Straight forward approach
GM deploys a public-private key pair for each group GM
PubG2 PubG3 PriG1
PubG1 PubG2 PriG3
PubG1 PubG3 PriG2
EPubG2(M)
EPriG1(M) G1 G2 G3
Key Distribution and Update for Secure Inter-group Multicast Communication

6. Straight forward approach
Three major disadvantages
The public-private key encryption involves exponential computation
Not efficient for a wireless node
The GM will be overwhelmed by the computation overhead for generating secure public-private key pairs when a group changes
An attacker can easily impersonate another node
Since the public keys are known to every node
Key Distribution and Update for Secure Inter-group Multicast Communication

7. New approach
Symmetric keys are used to protect the multicast traffic in intra-group communication
Polynomials are adopted to determine the keys to protect inter-group communication
Flat tables are adopted to distribute keys via broadcast when a group changes
Key Distribution and Update for Secure Inter-group Multicast Communication

8. Secure group communication
Intra-group communication GM
EKi-GM(K2)
EKj-GM(K2)
EK2(M) i j
EK2(M)
EKk-GM(K2) k G2
Ki-GM - pairwise key shared between node i and the GM
K group key shared by members of G2
Key Distribution and Update for Secure Inter-group Multicast Communication

9. Secure group communication
Inter-group communication GM
h12(x) h13(x) h21(j) h31(j)
h21(x) h23(x) h12(i) h32(i)
h31(x) h32(x) h13(k) h23(k)
Dh21(j)(Eh21(j)(M)) j i k
Eh21(j)(M) G1 G2 G3
h(x) - t-degree polynomial to determine the keys for decrypting the multicast traffic from other group
h(i) - personal key share to encrypt multicast traffic sent to the other group
Key Distribution and Update for Secure Inter-group Multicast Communication

10. Secure group communication
Secret keys held by node i in group G2 and their usage
Key Distribution and Update for Secure Inter-group Multicast Communication

11. Secure group communication
Secret key refreshment using the flat table
Flat table
Consists of 2r keys
r : the number of bits that are required to represent a node ID (r=┌log2n┐)
E.g., (z1.0, z1.1, z2.0, z2.1, … , zr.0, zr.1)
Every group has its own flat table
Every node has a set of keys in the flat table for its group
E.g., If r=4, a node ID with 10 can be represented as (1010)2
Flat table : (z1.0, z1.1, z2.0, z2.1, z3.0, z3.1, z4.0, z4.1)
The node has a set of keys (z1.1, z2.0, z3.1, z4.0)
Every pair of nodes in the same group must have at least one different key
Because every node has a unique ID
E.g., a node ID with 10 has a set of keys (z1.1, z2.0, z3.1, z4.0) a node ID with 11 has a set of keys (z1.1, z2.0, z3.1, z4.1)
Key Distribution and Update for Secure Inter-group Multicast Communication

12. Secure group communication
Secret key refreshment (Cont’d)
The flat table has brought two features
Only one node in a group can decrypt the message
Node i will have the keys (z1.i1, z1.i2, z2.i3, z2.i4, … , zr.ir)
can be decrypt by only node I
All the nodes but one node can decrypt the message
can be decrypt by all the nodes but node i
Key Distribution and Update for Secure Inter-group Multicast Communication

13. Key update during group changes
Group joining operations GM
EK1(K’1)
EK1(K’1) a b i
EK1(K’1) c G1
Step1. Update group key K1
Key Distribution and Update for Secure Inter-group Multicast Communication

14. Key update during group changes
Group joining operations GM
M : M M a b i M c G1
Step2. Update the new flat table for group G1
Key Distribution and Update for Secure Inter-group Multicast Communication

15. Key update during group changes
Group joining operations GM
M :
EK1(h’12(x), h’13(x)) M M a b i M c G1
Step3. Update the polynomials for inter-group communication
Key Distribution and Update for Secure Inter-group Multicast Communication

16. Key update during group changes
Group joining operations GM
EK1-GM(K’1, h’12(x), h’13(x), z’1.i1,…z’r.ir) a b i c G1
Step4. GM distributes the keys to node i
Key Distribution and Update for Secure Inter-group Multicast Communication

17. Key update during group changes
Group leaving operations GM
M : M M M M a b i c G2
Step1. Update group key K2
Key Distribution and Update for Secure Inter-group Multicast Communication

18. Key update during group changes
Group leaving operations GM
M : M M M M a b i c G2
Step2. Update the new flat table for group G2
Key Distribution and Update for Secure Inter-group Multicast Communication

19. Key update during group changes
Group leaving operations GM
M :
EK’2(h’21(x), h’23(x)) M M M M a b i c G2
Step3. Update the polynomials for inter-group communication
Key Distribution and Update for Secure Inter-group Multicast Communication

20. Discussions
Overhead
Compared to the group changes, the encryption and decryption of the traffics happen much more frequently
Additional transmission overhead for key refreshment is totally paid off
The adoption of polynomials enables the distribution of personal key shares
Difficult for an attacker to impersonate another node
When a node changes its group, new keys must be established by the group manager
Much efficient to choose several t-polynomials
Key Distribution and Update for Secure Inter-group Multicast Communication

21. Conclusions
Adopts polynomials to support the distribution of personal key shares
Employ flat tables to achieve efficient key refreshment
Reduces the computation overhead to process the packets
Becomes more difficult for an attacker to impersonate another node
Key Distribution and Update for Secure Inter-group Multicast Communication

22. Question?
Key Distribution and Update for Secure Inter-group Multicast Communication