Abstract:
This thesis proposes to implement a distributed solution for multiparty Voice
conferencing over peer-to-peer (P2P) mobile networks. The proposed system will take
the benefit of all the peers connected without using the centralized server. This system
will be fully distributed and all the audio streams are mixed on nodes and then
distributed to active members of P2P group. Peers are the nodes present at particular
instance of time in the conference. These peers are heterogeneous having different
bandwidths available, processing power, memory etc. The system will dynamically
form a topology according to current network. The algorithm will compute each peer’s
cumulative power depending on all the factors and build the matrix with the peer having
greatest cumulative power as the center node. However, the matrix topology does not
remain fixed during the session, which is dynamically changed. So at each time the
matrix topology will tends to have peers with higher power in the center and peers with
low power will be accumulated on the edges. The power of peers decreases from the
center of the row or column to the ends. This helps to distribute the load according to
the heterogeneity factors and thus brings the efficiency in distributed voice
conferencing system.
