Where Did We Diverge on Convergence?

As discussed in the first installment of this blog, reaching a common understanding as to what exactly is “convergence” can be a complicated matter. Catch phrases such as triple play and quad play appear to have become a must-have in the arsenal of the telecom marketing and sales personnel on either the selling or buying ends. Triple play has been used most widely to refer to a converged service offering of voice, data and video services. When wireless services are added to the mix, then the talk shifts to quadplay. This can be rather confusing when one considers that cellular phones, by a wide margin the bulk of the wireless equipment, are used primarily for voice communications.

In an attempt to stay away from the fray, I will instead divide electronic communications into the following three categories: Conversational, (tele)-entertainment and data.

Conversational communications essentially mimic traditional human interaction and normally have the highest QoS requirements due to their real-time nature and their high information content. Conversational communications are by definition bi-directional and include voice only, and, voice plus video (video calling/conferencing).

Entertainment-related communications have varying real-time requirements depending on their inherent level of interactivity and the immediacy of the content. These range from non-real-time audio/video (A/V), to A/V-on-demand, near A/V-on-demand, broadcast A/V, and such things as remote gaming.

Finally, in the context of this classification, data services are considered as having typically the lowest real-time requirements.

Before diving into how these services should be implemented within a future converged network, it is worthwhile first to revisit how they have been provided traditionally.

Contrary to popular belief, data services were the precursor of all telecommunications with the advent of the telegraph in 1860, which was the first digitized electronic service. Yes, I am being tedious, but correct. Voice communications followed, in 1876, with the advent of the telephone. Last, tele-entertainment was introduced with the advent of the broadcast radio around 1910 and broadcast television in the 1930s.

Of these early services, voice typically had the highest real-time and quality requirements, followed by the broadcast A/V services. For the average customer, the telegraph emulated a traditional shared postal service with just the element of speed added.

With the exception of the telegraph, as pointed out above, voice services and broadcast entertainment services relied on analog transmission techniques that were not well suited to time-division multiplexing, but fared well when channelized using frequency-division multiplexing. Fixed bandwidths were allocated for voice, audio and video transmissions, and this arrangement was maintained for both terrestrial and cable-based broadcasts.

Digitization was introduced for voice communications in the 70s, and resulted in the time-division multiplexing model so prevalent in the PSTN. However, channelization, using a fixed dedicated bandwidth per connection, also was retained because of technological and QoS reasons.

The computer revolution of the 80s ushered in the World Wide Web in the 90s. This essentially non-real-time service relied on packetization and followed the shared postal and telegraph model at the ISP level and above. Packets were suited ideally to transport the IP content and offered the bandwidth sharing needed to keep costs low and make Internet access affordable to the masses. Initially, fixed low-bandwidth connections linked the users to the ISPs, using the existing voice network based on the omnipresent Class 5 COs.

So, what we have then, at least in the access and metro areas, are principally just two networks: a voice/data network and a broadcast entertainment network.

What happened next may be considered the first divergence. The telecom service providers’ venerable Class 5 COs were stressed by this new data traffic, since they were not designed to support the long call holding times of the Web surfers. In many cases, expanding these Class 5 switches was not even a reasonable option, since most of them were engineered primarily to support a relatively high subscriber-to-port ratio. Consequently, in conjunction with the advent of DSL, which further aggravated the bandwidth issues, the telecom service providers chose to reroute the data traffic around their original CO equipment, and so the DSLAM was born.

The only converged element, at this point, is pretty much strictly the physical connection from the CO to the customer Network Interface Device (NID). Within the customer premise, the DSL splitter/modem combination creates separate voice and data networks. When arriving at the CO, the DSLAM again separates the data stream into separate voice and data networks.

The “convergent” option that could have been used for the COs at the time would have been the wide-scale deployment of ATM switches to replace these COs. Why this did not happen will be the topic of the next installment of this blog.

Serge Fourcand is a principal engineer with Huawei Technologies USA. He is an accomplished product manager and system architect with extensive multimedia experience in both consumer and telecom industry segments, including end-consumer multimedia products and associated controls, advanced multimedia distribution, speech processing, IP-based video and audio transport, digital signal processing, Ethernet Layer 2 transport and switching, and end-to-end telecommunications technology. Prior to Huawei, Fourcand worked as a product manager for AMX Corp., senior hardware architect at Metro-Optix, a system architect for Broadband Gateways and Alcatel USA’s Switch Products Division, and has been a member of the technical staff of Siemens AG in Munich, Germany.

Share this article: Email, Slashdot, Digg, Del.icio.us, Yahoo!MyWeb, Windows Live Favorites, Furl
Add this article feed to: RSS, My Yahoo, Newsgator, Bloglines