| The phenomenal growth of VoIP is attributed to its cost saving and functional advantages over the traditional telephone line. Most of the protection for voice in today's network comes from over-provisioning, which is a viable option for an enterprise network, but may not be suited for public network. In public networks, the inherent unreliability of the router and network congestion often leads to unacceptable performance. Voice traffic from telephone conversation cannot be achieved reliably if the links have large packet loss or the mount-to-ear delays.; The network architecture for voice has major role to play in deciding the kind of protection is needed, e.g. the protection methods for central office based VoIP deployment will be very different from end user based deployment. The former could choose for SONET based protection, while the latter has to rely on IP based protection. For this reason, various VoIP deployments are discussed and general background on voice transportation over the IP is discussed. The review also includes a discussion on various VoIP components, and the factors that affect voice quality.; There are several protection mechanisms to achieve this reliability, and can be used at different protocol layer. This thesis focuses on IP layer protection. Such protection is achieved by MPLS or Differentiated Services. In a multi-homed network, additional protection can be provides by packet duplication. These techniques can be combined together. Unlike MPLS, Differentiated Services are simple to deploy and do not require major network upgrade.; The simulations were conducted for multi-homed networks with voice protected by premium class of Differentiated Services and packet duplication. The results suggest the greatest benefit is achieved when these two are used together. The simulations also show that smaller voice packets have better delay and jitter behaviour. However, in such packet the protocol overheads are large, but they are better choice if used along with the header compression. |
