Information Technology Reference
In-Depth Information
KHT key
update
Group A
KHT
Cache
Manager
LKH key
update
User 1
Add/Remove user
LKH
Key
Manager
User 2
Fig. 1.
Overview of the key management strategy
by the LKH-KM. During this pre-processing, the KHT-CM uses information
from previous key updates, so that users are more likely to recover using the
transformed update if they have been off-line. Moreover, to minimize the increase
in size of the update, the cache prunes contextual information that is likely to
be already known by group members. However, this means that on some rare
occasions, even with this extra information a user may still not recover. In this
case he will re-register directly with the LKH-KM.
The proposed design makes the KHT caching as “transparent” as possible.
First, the KHT-CM neither mediates authenticated interactions between users
and the LKH-KM, nor needs to decrypt key updates from the KM. Therefore,
if it gets compromised it could perform a denial of service attack, but not an
attack that compromises the confidentiality of the group keys. Moreover, key
information updates from the LKH-KM use the same format as the ones obtained
from the cache; this makes possible to add single or multiple levels of caching
without modifying clients of an existing LKH implementation. Finally, the LKH-
KM does not need to be aware that its content is being cached because the only
interaction with the KHT-CM is to “forward” authenticated messages to be
broadcast.
KHT offers robustness for group rekeying with off-line members when it is
undesirable to either rekey every time a member becomes off-line or trigger an
expensive recovery action when back on-line. This is the case when members
are off-line frequently, the group membership is very large, or it is impossible to
accurately determine who is on-line at a particular time. Well-suited applications
include wireless secure group communications among low power devices, loosely
coupled secure groups that only synchronize keys from time to time due to
connectivity restrictions, and anonymous subscription services, such as AGCD,
the focus of our simulation efforts in Section 5.
3.2
Updating the KHT
A KHT is implemented by a simple data structure that summarizes all the public
information that the KM multicasts, so that any group member can compute
