VoIP Does Not Have To Be POTS Over IP!
Its Time to Take IP Telephony Beyond Simple Replication of Traditional Telephony Service

Most of todays VoIP services merely offer conventional telephony at lower prices. That market is very volatile, since more and more players must share a decreasing revenue stream. On the other hand, IP technology and Session Initiation Protocol (SIP) allow for much more than replication of old telephony. Soon, live person-to-person IP communication, including new and extended services and global IP-to-IP connectivity must be offered. The success of Skype has killed several myths in this area, but future global telephony systems must be based on open standards (SIP), allow for a wide range of applications, and be extendable.

Lowering Cost, but Thereafter...

The driving factor for most of todays VoIP offerings is lower cost for a conventional telephony service. However, using new, advanced technology for simply reducing prices, in a market that already is covered by existing services, will not allow stable and healthy market players. Instead, the total amount of money earned will decrease and will have to be shared by a growing number of market players. Costs do need to be reduced, but customers should also expect innovative new products and features.

While IP telephony technology and the SIP standard have the capacity for new functionality and applications, current VoIP deployments seldom offer more than plain voice communication with telephony quality. VoIP service providers even boost quality as good as POTS (Plain Old Telephony Service) or PSTN (Public Switched Telephone Network), without considering that voice limited to 3kHz has been around for 50-100 years and is far from todays technological capabilities. Compared to radio and TV, todays VoIP is like introducing AM quality digital radio, disregarding that technology long ago enabled both full bandwidth sound and TV.

Stranger still is that VoIP providers do not connect to other VoIP providers customers over IP; instead, they squeeze the traffic through PSTN lines. This not only increases cost, but it limits the functionality for the IP user as well. Features like better sound quality through higher bandwidth, video, presence, and instant messagingall supported by the SIP standardare made unavailable to the users becuase the calls are being routed through the PSTN.

Comparing VoIP to the development of the wireless phone services (e.g., GSM and 3G) there is wide contrast. Wireless phones have not only given us mobility, but also text messaging, pictures, and video. Wireless handsets also have become more and more versatile, with the introduction of various multimedia capabilities and products, including PDAs and cameras. This has caused the wireless telephony market to grow rapidly and experience a healthy revenue stream.

The huge, but inflexible, PSTN infrastructure built over the years has prohibited substantial developments in call quality and functionality for the traditional telephone service. When VoIP services are being built, it is, therefore, important to not merely replicate POTS. Instead, the flexibility, capabilities, and advantages of IP, the Internet, and the SIP standard must be maximized and offered to the customers. This will allow usersanyone with an Internet connectionto communicate globally, person-to-person, using a variety of media and new functions and applications.

SIP-based live person-to-person communication has the potential to become the next commonly used application on the Internet, after email and the Web, and to create a healthy and growing market. In other words, VoIP technology must be feature-rich and cost-efficient, so as to outshine both PSTN and the wireless industry.

Why Replicate PSTN and POTS?

It is unfortunate to only offer POTS-like telephony, replicating both the old telephony service and the old telephone network, when VoIP services are being built. Operators actually invest in soft switches and Session Border Controllers (SBCs) that, most often, are used to do just that. VoIP is being built as closed islands and traffic is both limited to 3kHz voice and fed into the PSTN. So, even if both endpoints are SIP clients, but belong to different islands, they will unnecessarily connect via the PSTN, rather than directly via IP.

Recently, VoIP providers have begun to consider the potential advantages of peering with one another via IP. However, that is an artificially introduced need emanating from the VoIP islands being built. SIP servers following the standard do talk directly to each other and to their clients over the Internet, just like email does! Today, no one would even consider sending email via the telefax service, limiting it to only pixel transmission and getting the old slow delivery time and higher cost. That was the method used in the early 90s, before we started using the SMTP Internet standard. But that is often how VoIP is deployed today, even though we have the common Internet and SIP Internet standard!

Skype Has Killed Some Myths

Reasons abound for building VoIP islands connected to the old PSTN, instead of building an open global VoIP service over the Internet. The most common are to guarantee quality of service (QoS) and that VoIP cannot scale without the structure of the PSTN. The success of Skype has certainly aided in killing those myths! The Internet is actually capable of much higher quality than the PSTN and, using the SIP standard, scaling is done via DNS, the Domain Name Services that has allowed unlimited expansion of both email and the Web.

Skype also has shown that it is possible to get extremely rapid acceptance and penetration without using telephone numbers or hardware phones. VoIP providers that curently do not give users a proper SIP address (in addition to a telephone number), should consider what email would have been if we had been forced to use fax numbers for addressing.

The NAT and Firewall Problem

Today, more and more people have good broadband Internet connections both at home and in the office. There is also the presence of good SIP servers and SIP clientsin the form of SIP hardware phones and soft PC clients. Considering that, why isnt global SIP-based person-to-person communication already as commonly used as e-mail?

One major obstacle is the NAT (Network Address Translation) and firewall problemall too frequently, protocols for connecting directly to individual users simply do not pass firewalls. NAT ultimately means that users on a LAN cannot be reached via IP addresses and firewalls are used to block unknown traffic. The real problem lies in that many firewalls still lack proper support for SIP, the standardized Internet protocol for live person-to-person communication. Additionally there is a large installed base of totally SIP unaware firewalls.

Skype can handle the firewall problem is most cases, but unfortunately does it using, at best, underhanded methods. Skype uses a closed unpublished protocol and runs an application on each PC, over which only Skype has full control. That application helps penetrate firewalls from inside private LANs. If Skype cannot punch holes in a tight firewall for its signaling and media, it masks itself as HTTP, for which firewalls are open to allow surfing. These methods remove the control from the firewall and the firewall manager and are not acceptable to security-aware enterprises.

To allow global SIP usage before all firewalls have proper SIP support, several workarounds have been proposed. STUN, TURN, and ICE are methods where the SIP client, together with the servers on the Internet, try to punch holes in NAT and firewalls. However, these methods rely on guesswork of the firewall and NAT behavior and wont work in all cases and certainly will not be particularly effective or reliable. They also move control of what should pass the firewall from the firewall manager to the clients on the network in the same unacceptable way that Skype does.

Another method, which seems to have a higher success rate, is far-end NAT traversal, where the operator equipment tries to punch holes in the customer firewall. In this case, firewall control is moved to the operator and only helps in accessing that operators services, not SIP services in general. For this to work, however, the firewall has to be sufficiently open.

The most general solution is, of course, to solve the problem at its sourcein the NAT/firewall itself. Firewalls with good, general SIP support today exist from a few vendors, for both residential and enterprise usage.

For those unwilling to replace an existing firewall, a secondary SIP-enabling device, such as the SIParator or an SBC, can be parallelled with the existing resource to create the same functionality as a good SIP-aware firewall.

The lack of functional SIP support in the many popular firewalls sold today is a ticking time bomb; most of these firewalls will not even be upgradeable to handle SIP.

To be part of the accelerating SIP user community, which now includes many IP PBXs, it is important for networks to be prepared. To have universal connectivity across the Internet, NATs and Firewalls need to be SIP-capable, which, unfortunately, is presently uncommon. IT

Karl Erik Sthl, is President of Intertex Data AB, Sweden. For more information, please visit www.intertex.se.

If you are interested in purchasing reprints of this article (in either print or PDF format), please visit Reprint Management Services online at www.reprintbuyer.com or contact a representative via e-mail at [email protected] or by phone at 800-290-5460.