Page 3
Page 3 |
|
Increasingly, fiber cables are replacing copper for T1/E1 and PRI ISDN delivery. From the user perspective, nothing changes – the Telco will provide the boxes to convert the optical connection to the standard copper T1-style RJ jacks and all your gear connects as if the delivery were on copper pairs.
If you are a really big user, or are simply in the right place, you may get your feed directly from a tap on the Telco’s SONET (Synchronous Optical NETwork) Ring. This is fiber writ large: an OC-1 has a 51 Mbit/s data rate, while an OC-3 can handle a whopping 155 Mbit/s. Because SONET uses an add/drop scheme, you can break out (and pay for) only the capacity you need in T1-sized chunks.
Fig.2: SONET Ring connection may be the ultimate hook-up.
There is yet another digital telephone connection possibility on the horizon: Voice Over IP, or VOIP. All of the systems described so far are so-called Circuit-Switched connections over channelized bit pipes. That is, the calls are handled as continuous streams of bits at a 64 kbit/s rate. While a call is connected, the channel is nailed-up and dedicated to this one call. The Internet uses a fundamentally different way to convey its bits: packets. These are groups of bits organized according to Internet Protocol (IP) that include a header containing the source and destination addresses. Using these addresses, routers, placed at various nodes, direct the packet to the specified end point. Voice signals can be encoded and packetized for transmission over the Internet – bits are bits, after all. IP packets offer flexibility compared to circuit channels, but at a cost.
Current VOIP has a deserved bad reputation for quality. The main problem results from the current state of the public Internet, where there is no guaranteed bandwidth for a call: all packets on a segment contend for capacity, and it is often the case that there is more demand than space – making for choppy audio. But there is no technical reason why bandwidth can’t be assigned and assured, and next-generation and private IP networks will almost certainly support this service. The second problem is poor audio quality – even when there is enough bandwidth – caused by the compression applied to conserve bits. Even the core 64 kbit/s in a standard voice channel is too much for modem connections to handle, and packetization adds a lot of overhead – as much as 100%. So VOIP designers must use aggressive compression to reduce bit rate. Typical is 9 to 16 kbit/s, and this doesn’t do much for the audio! But future networks will have plenty of capacity and it should not be necessary to compress to the extreme.
The main appeal of Internet phone calls now is cost, but it is not clear if high-quality VOIP will preserve this advantage. Packets are less efficient from the perspective of bit consumption, owing to the addition of headers, checksums, etc. This can be made up with compression, but to the detriment of audio quality. If providers give quality guarantees comparable to traditional phone service, they will consume more bits per call than the current method, so might need to charge customers at close to circuit-switched rates.
Delay is another issue. Because it is tremendously inefficient to put only a single audio sample in a packet, multiple samples are buffered, accumulated, and packed into each packet. This requires buffering, which causes delay in the audio. Then, more buffering is added to cover network delays.
Nevertheless, IP’s flexibility is likely to mean that it will eventually be widely deployed for voice, owing to the advantages it brings:
Common cables, switches, etc. can support both voice and data.
IP routers are inherently simpler and will be cheaper than circuit switched CO equipment.
Routers and associated networks are more readily scalable.
A wide range of media types and different audio quality levels are easily possible simultaneously on a single network.
The computer industry has tremendous financial clout and R&D resources with which to challenge traditional telecom vendors. Cost per bit for transmission is steadily and dramatically falling, making IP’s inefficiency less a problem, and its flexibility more an appeal. Some telecom pundits say “bet on bandwidth,” comparing this to the bets on CPU speed increases that created winners in the last computer industry round. Today this is all very interesting to observe, but pretty much academic for professional users. Until the problems are sorted out and serious service levels are available, VOIP is not an option for broadcast applications.
• 
