Single Leave. In case of single leave, key path from the leaving member to the root of the tree must be updated. If the leaving member is group controller (GC) then choose some other member as the group controller but no need to change other subgroup controller.Leave operation in proposed protocol is very similar to that in TGDH [11] .Suppose that we have n members in the group and assume that member Md leaves the group. First, each member updates its key tree by deleting the leaf node corresponding to Md. If Md have only one sibling, the former sibling of Md is promoted to replace Md’s parent node. Leave operation in proposed protocol is very similar to that in TGDH [11] the sponsor generates a new key share, computes all {key & blind key contribution with others siblings} on the key-path up to the root, and broadcasts the
Single Leave. In case of single leave, key path from the leaving member to the root of the tree must be updated. If the leaving member is group controller (GC) then choose some other member as the group controller but no need to change other subgroup controller.Leave operation in proposed protocol is very similar to that in TGDH [11] .Suppose that we have n members in the group and assume that member Md leaves the group. First, each member updates its key tree by deleting the leaf node corresponding to Md. If Md have only one sibling, 4.3.2 Mass Leave: If the no. of leaving members are very large then it is better to reinitialize the tree with other than the leaving members. So if bJe the total no. of members in initial group and Iis the no. of leaving members from current group then The message cost of the mass leave operation is: Ru: Æ? ? 5; S Total no. of point multiplication is: 9: Æ? ? 5;E D : JF I ; Note that after numerous group operations like join, leave, mass join etc. the resultant key tree becomes quit unbalanced. So as suggested in [27] when the key tree reaches a certain imbalance point, we should rebalance the tree and treat the rebalance operation as the group operation required. The imbalance point can vary depending on network and efficiency requirement. The proposed paper does not provide any new rebalancing scheme however protocol can use any scheme discussed in [27]. Also sometimes it is more suitable to reinitialize the tree instead of rebalancing. 5. AUTHENTICATION Although the proposed protocol is based on ceullrivpetic Diffie Xxxxxxx which is secure and not eabsryeatko, the entire system is vulnerable if the keys arseecnuortely distributed. Therefore, we should impleamneanutthentication algorithm ni the protocol. There asreveral ways to authenticate a group key exchange, sauschcentralized W K H I R U P H U V L E O L Q J R I 0 G L V S U R P R W H node. Leave operation in proposed protocol is very similar to that in TGDH [11] the sponsor generates a new key share, computes all {key & blind key contribution with others siblings} on the key-path up to the root, and broadcasts the authentication, implicit authenticatioann, d pairwise authentication [3, 4, , 911, 16 ]. In all operations of the protocol any of the above authetniotincaschemes canubseed. The advantages and limtioitnas of dffierent authentication schemes are discusinseCdCEGK [27].
