Peer-to-peer (P2P) mobile VoIP

Three years ago, peer-to-peer technology was all the rage made popular by Skype and file sharing applications such as KaZaA, Shareaza, and BitTorrent. I too caught the P2P bug and together with two other co-founders developed a P2P-based distributed telephone switch. Coming from a telecom background in mobile network equipment and Internet Multi-media Subsystem (IMS), we thought the use of P2P for node addressing and Session Initiation Protocol (SIP) for call signaling would make for a powerful and low cost disaggregated telephone soft-switch. We experimented with several P2P protocols including Sun's JXTA, Gnutella2 and FastTrack but eventually ended up with a beta product based on the more structured Pastry P2P design. The rationale was that to have a P2P network that can match the performance of traditional telephone switches, we needed a P2P protocol that is more real-time deterministic and one that can be made practical on limited resource end-user device such as mass market mobile phones.

We never released the product but we did run an emulated large P2P network for a while on many servers that indeed show a pretty large functional Pastry ring. And in the process of building a technology that can bridge one telecom service to another, we actually ended up with more technology on the server-side than client-side P2P technology. Part of the reason was that SIP, while mostly an end-to-end "peer-to-peer" call signaling protocol, is typically implemented as client-server solutions. Another reason was that for mobile devices with limited memory, limited computing, limited bandwidth, and limited battery resources, P2P technology is less effective compared to client-server technologies because there are not a lot of resources or too costly to share compared to P2P client applications running on PCs and fixed-line broadband connections such as ADSL and PacketCable.

So is there a future in P2P mobile VoIP? I think so but likely not based on conventional P2P designs in the near term. With increasing memory, computing, and bandwidth resources on WindowsMobile, Symbian, RIM BlackBerry, embedded Linux, and Palm smart phones, it is becoming more and more practical to run P2P applications especially on Wi-Fi dual mode devices. However, battery consumption and bandwidth requirements (it takes about 6 times more bandwidth for native VoIP voice calls compared to circuit switched calls) will continue to be challenges. Also, P2P client applications on these mobile devices are typically proxied by servers in the core network which makes the technology not all that different from a SIP user agent running on a mobile phone such as the Nokia E61 controlled by a SIP server. Considering that a SIP server on a low cost server machine can easily serve hundreds of thousands SIP clients running on mobile phones, there are far less cost benefits associated with traditional P2P designs. P2P-based multi-media file sharing from the mobile phone anyone?

Colin - EQO Founder & CTO