- In The Beginning...
- Power To The People!
- The Network Goes Digital
- Business Takes Control
- The Digital Local Loop
- Digital Access Lines
In order to fully appreciate the current state of data communications technology, it is important to have an understanding of the Public Switched Telephone System (PSTS), also known as the Plain Old Telephone System (POTS). The construction of the basic telephone has not significantly changed since the original system demonstrated by Alexander Graham Bell. On a standard telephone set (Type 500), we still speak into a carbon microphone and listen to a small speaker. The technology behind this simple mechanism has changed greatly over time, however, and continues to evolve to this day.
In the original telephone system implementation, a telephone located in the customer premises (Customer Premises Equipment, or CPE) was connected by two copper wires to a local Central Office (CO). The wiring from the CO to the customer premises is known as the subscriber loop, or local loop. The various Cos were interconnected by trunk lines to allow calls from a telephone on one CO to be placed to a telephone on a different CO. The local loop was, and mostly still is, an analog system: voices are sent and received as Amplitude Modulated (AM) signals on the local loop.
Originally, in order to place a call, the hook switch would be "flashed" in order to attract the attention of an operator in the CO. The operator would then connect to the line, indicate by the flashing light, by inserting a "phone plug" on a switchboard to the line. The standard phone plug is show in Figure 1; note that, to this day, the red and green wires on the local loop are called ring and tip. The subscriber would then ask to be connected to the party to which they wished to speak. The operator (possibly with the assistance of other operators in other Cos) would then establish the connection between the two parties.
To indicate an incoming call, a ringer would be connected to the destination subscriber loop. The ringer frequency of 20 Hz would (generally) have a duty cycle of two seconds on, four seconds off. A busy line would be indicated to the caller by the operator. While this system was adequate for a small number of telephones, as the popularity of the device exploded, the American Telephone & Telegraph company (AT&T) realized that it would eventually require more operators than the entire population of the United States. This led to the development of direct dialing.
Direct dialing required a mechanism for the subscriber to specify the destination and a mechanism to automatically switch the call. The standard telephone dialer effectively flashes the hook switch at a rate of 10 pulses per second. Dialing the numbers 1 though 9 generated the same number of dialing pulses while dialing 0 generated 10 pulses. The original switches were electro-mechanical in nature. The standard rotary switch consists of a shaft which is lifted by one stepper motor and rotated by another. The number of steps taken corresponded to the number dialed. Dialing 43, for instance would cause the rotor to be lifted four steps and then rotated three.
Since operators were not involved in call placement any more, there had to be a means to indicate to the caller that the destination loop was not available. The busy signal is used for this purpose. The frequency and duty cycle of the busy signal was originally quite variable, but it is now specified as Ref. The power required by switch banks using rotary switches was enormous. AT&T developed fully electronic switches to replace the cumbersome electro-mechanical switches. Please note, however, that there are still exchanges in existance which use the old rotary switches.
Going from electro-mechanical to fully electronic switching also permitted the telephone companies to provide an alternate dialing mechanism. Using two sets of tones for each number on a telephone dial, the TouchTone (TM) dialing mechanism was made available to customers. It permitted faster dialing and was generated electronically, permitting the elimination of some of the electro-mechanical parts in a standard telephone. This mechanism is also known as Dual Tone Modulated Frequency (DTMF) dialing.
The newer electronic switches provided ever more services to the consumer. Within most areas of North America it is possible to subscribe to a number of special services. These include the ability to receive an indication of an incoming call while you are talking on the phone, customer-controlled call forwarding, display of the originating telephone number of an incoming call, and others. Please note, however, that these services are still being provided on what is still an analog line which will only permit one connection at a time.
While many developments improved the service provided to the customer on the local loop, the telephone companies were also making significant advances in the rest of the infrastructure. It quickly became apparent that using a pair of wires for each call was inefficient. It became essential to improve the utilization of installed trunk lines since adding more trunk lines could be a very expensive proposition (especially in built-up areas where it would be necessary to dig up streets to replace or supplement trunk capacity.)
The solution was to multiplex a number of conversations on a single wire pair. This techique is known as Time Division Multiplexing (TDM) as each conversation is allocated a time-slice in which to send information. In order to use this technique, however, the telephone signal must be converted from an analog to a digital format at the origin and converted back to analog at the destination. This is performed by specialized chips known as Coder/Decoders (CODECs).
The telephone CODECs use a mechanism known as Pulse Code Modulation to encode the analog signal as a stream of binary digits (bits). This technique samples the analog signal 8000 times per second (every 125
msecs) and encodes the sound intensity as seven bits according to an algorithm which gives more weight to lower instensity sounds. For each sample, an eight-bit byte is generated, containing seven bits from the aforementioned algorithm and one bit for use by the telephone companies. This generates an aggregate data rate of 64000 bits/sec.Depending on the speed of the circuit, itself somewhat dependant on the circuit length, a number of voice channels can be multiplexed. A link running at a speed of 1.544 Mbits/sec (DS1 speed) can carry 24 conversations simultaneously. Ref. contains information about the standard trunk line speeds and the number of voice channels which the link can carry. As an aside, note that the 1.544 Mbits/sec speed can be maintained for a distance of up to 2.4 Kilometers from the CO. This fact will be important in later discussions.
The telephones companies have been able to wring even more performance from their available trunks. Instead of using a 64Kbit/sec channel for a single conversation, it is possible to have two or even more conversations on the same channel. The technique known as Adaptive Differential Pulse Code Modulation (ADPCM) is similar to regular PCM but each sample (taken at 125
msec intervals) records the difference in sound intensity between the current and previous samples. This difference is encoded in only four bits, resulting in a bandwidth of 32 Kbits/sec. Two such ADPCM channels can be carried on a regular 64 Kbit/sec channel.The special requirements of businesses forced the telephone companies to provide additional services. One of the first of these was the Central Exchange (Centrex) service which used the facilities of the CO. Eventually, however, businesses wanted to have more control over the telephone services provided to their employees and customers. This led to the development of the Private Branch Exchange (PBX). The PBX at a customer premises is connected to the CO through trunk lines. While these trunks were originally analog, as the telephone companies moved to digital so did the PBX makers. For the last few years, all PBXs have been designed to use digital trunks.
When PBXs used analog trunks, conversion of the signal from analog to digital was still performed in the CO. With the use of digital trunks, the CODEC function was provided by the PBX.These days, in addition to using digital trunks from the PBX to the CO, most of the PBX systems today use digital telephone sets as well. This permits additional functionality to be integrated into the handset but requires each handset to be equipped with a CODEC, moving digital communications all the way to thedesktop. Figure 2 shows the general layout of the modern telephone system components.
Now that digital communications technology has come so far, the only remaining analog component of the modern telephone system is the local loop provided to individual residential customers. The analog loop technology is very mature and therefore inexpensive, but does not begin to address the new technological requirements. Dial-up data communications still typically uses a modulator/demodulator (modem) to convert a digital data stream to an analog signal which can be carried over the PSTS. Even with claims of speeds up to 38.4 Kbaud, actual throughput on analog lines is slow compared to the speeds of modern computer devices.
Integrated Services Digital Network (ISDN) is a standard which will replace the last vestige of analog signalling, the local loop, with a digital link. Running at a speed of 192 Kbits/second, the local loop can carry two 64 Kbits/second bearer channels (B-channels) and a separate 16 Kbits/second signalling channel (D-channel). The ISDN standard specifies a number of different interfaces and hardware/software components. This is show in Figure 3. Note that calls outside the local CO might be subjected to further compression by the interexchange carrier (remember the discussion on ADPCM.) Be sure to look for a guarantee that your telecom provider will carry your calls only on 64 Kbit/sec "clear channels".
The U-interface is the two-wire local loop. The NT1 device is either a stand-alone unit or can be integrated with the NT2 device. The NT1 device translates between the protocols used at the U and T interfaces. The NT2 device permits access to the T interface protocol to up to eight TEs on a passive S bus. TE1 devices are ISDN-aware and connect directly to the NT2 through the use of the protocol. TE2 devices, such as older terminals, communicating using RS-232, RS-449, V.35 or X.21, connect to the NT2 through a TA, or terminal adapter. Analog devices, such as telephones and modems, can still be attached to the line through the use of a CODEC.
The D-channel is used for call setup and call mangement. It can also be used for low-speed access to a packet-switched network provided by the telcom supplier. Depending on the facilities requested at the time of ordering the line, you have the ability to manage more than one call at a time through Multiple Call Appearances. This will allow you to perform such tasks as take down an established B-channel call in order to accept another incoming call or to put calls on hold while taking the another call.
One of the most powerful features of the call notification in ISDN is the provision of the calling telephone number in the call setup request. With the appropriate interfaces, this calling number could be used as a lookup to a database which could then provide name and other information to an operator responding to the call. Also note that the services which the telephone companies would typically provide piece-meal at additional cost can all be configured using the facilities of ISDN. Even automatic call forwarding is simple to implement using ISDN.
The previous discussion of the digital local loop has focussed on what is known as the Basic Rate Access (BRA) or Basic Rate Interface (BRI). It provides 2B+D service on standard local loops. There are some restrictions regarding open taps and loading coils on the line, but removal of these impediments is a one-time task. There is another interface available called Primary Rate Access (PRA) or Primary Rate Interface (PRI). It provides 23B+D services on a 1.544 Mbit/sec channel. The PRA service can be carried over either copper wire or optical fibre.
The B-channels on PRA are identical to those on BRA. The D-channel on PRA, however, runs at 64 Kbits/sec rather than 16 Kbits/sec speed on BRA. While PRA is available on local loop wire pairs, it is typically only provided when the total length of the loop from the customer premises to the CO is less than approximately 2.4 Kilometers. This total length could be considerably greater than what might be considered sensible, due to the way that the telecoms route trunk lines. If your telecom provider needs to install fibre for PRA service, be prepared to pay a very high initial installation charge.
The original Ethernet was based on work done by Xerox at the Palo Alto Research Centre (PARC). It uses a coaxial cable backbone to which a number of nodes connect for the purposes of intercommunication. The connection to the coax is typically made with a Media Attachment Unit (MAU, or "vampire tap") which has a tapered pin conductor which makes contact with the central conductor of the coax. A clamp connects to the exposed metal braid of the coax. The transceiver interfaces to the node through the Attachment Unit Interface (AUI) cable.
Ethernet cable is a broadcast media to which each node has transmit and receive access. Communications over the coaxial cable occurs at a rate of 10 Mbits/sec. Access to the transmission media is controlled by the Carrier Sense Multiple Access/Collision Detection mechanism. A transmitter will not attempt to send data unless it detects no activity on the "ether". Once it begins transmission, it also listens to confirm that no other transmissions can be detected on the ether. If it does detect another transmitter operating, it will cease transmission for a period of time before attempting to send again.
Each receiver has access to all transmissions on the cable. Received packets which should be handled by the node are passed to higher-level protocols. Each Ethernet interface has a 48-bit hardware address assigned by the manufacturer. This mechanism ensures that every interface has an address unique in the world, eliminating at least one configuration step if the interface moves to a different network. Ethernet frames contain source and destination Media Access Control (MAC) addresses in the header of the packets sent across the network. MAC addresses are typically written as six groups of two hex digits each. The special address of hex FF.FF.FF.FF.FF.FF is used as a broadcast address which can be received by all nodes on the network.
The original Ethernet was the basis for the Institute of Electrical and Electronic Engineers (IEEE) 802.3 standard. While the Ethernet specification defined only a single transmission medium, the IEEE 802.3 standard permits a variety of different media. The specification of physical characteristics of the IEEE 802.3 communications medium is made in the form
The 10Base5 IEEE 802.3 specification is essentially the same as standard Ethernet, indicating a 10 Mbit/sec baseband medium with a maximum segment length of 500 meters (approximately 1640 feet.) This medium is commonly known as "thick-wire Ethernet". A less expensive alternative is what is known as 10Base2, which communicates over a thinner coaxial cable and interfaces through a BNC connector on the interface card. It is essential that the "thin-wire" coax is properly terminated at both ends.
The latest incarnation of IEE 802.3 media is known as 10BaseT. It uses a central hub/retimer with a connection to each node carried over unshielded twisted-pair (UTP) wiring with RJ45 connectors at each end. The maximum length of the UTP connection is typically 100 meters. In modern buildings, it is often possible to use preinstalled cabling to implement a 10BaseT Ethernet LAN, although older buildings will still need new cables laid. There are a number of companies making adapters which convert from AUI to 10BaseT, as well as companies which make interfaces with both BNC and RJ45 connectors.
IBM developed the token-ring network as a mechanism for connecting the original IBM PC’s together. The term comes from the fact that the network is implemented as a logical ring and write access to the ring is controlled by ownership of a "token". A node can only write data to the network when it is in possession of the token. The token is passed around the ring so that each node has an opportunity to write a packet to the network. Data transfer takes place in only one direction on the ring.
One of the advantages of the token-ring is that it ensures fair access to the transmission medium. Each node is guaranteed to be able able to transmit a data packet each time the token circulates on the ring. The implementation has an active monitor which controls existance of the token, ensuring that only one token circulates and effecting the regeneration of the token when it ceases to exist. Each node knows to pass the token once it receives the data which it transmitted. It passes the token immediately if it has no data to send.
Access to the ring is typically gained through a Media Access Unit (MAU). This unit contains a relay on each port, permitting the electrical disconnection of a port from the ring unless a phantom voltage is asserted by a connected adapter. MAUs also contains Ring-In and Ring-Out (RI and RO) ports which permit the construction of larger rings. Token-ring networks originally operated at a speed of 4 Mbits/sec but this has now been increased to 16 Mbits/sec.
Token-ring communications are not limited to PC-class computers. Token-ring Interface Controllers (TICs) are available for mainframe communications controllers such as the IBM 37xx Front-End Processors (FEPs). With the use of appropriate protocols, token-ring networks permit the simple connection of various device types to mainframes. While these network access protocols were initially designed to support the Systems Network Architecture (SNA), other developments permit the use of other protocols.
The IEEE 802.5 specification is based on the original Token Ring Network and is effectively identical. As a result, the term token-ring LAN is taken to mean both the original IBM implementation and the IEEE specification. As with the differences between the Ethernet and IEEE 802.3 standards, while the IBM token-ring standard specifies the communications medium the IEEE specification is not bound to a particular medium or even a topology.
The Fibre Distributed Data Interface specifies a LAN operating at 100 Mbits/sec over optical fibres using either a star- or ring-based architecture. Packet framing is similar to the Ethernet specification. The attributes which make FDDI desirable as a backbone networking product are listed in the sections which follow.
Data Transfer SpeedA major advantage of FDDI over ethernet and token-ring LANs is the speed of data transfers. While ethernet has a theoretical throughput of 10 Mbits/sec, in actual use the utilization of the available bandwidth rarely exceeds 50%. Token-ring networks can run at the full 16 Mbits/sec speed but the bandwidth is again not fully utilized as the transmitter waits until the send packet is received before handing the token to the next transmitter. Even though FDDI uses a variant of CSMA/CD to control access to the network, very high utilization of the available bandwidth of 100 Mbits/sec is possible.
The speed of FDDI permits the deployment of applications which would not be appropriate for other LANs. With FDDI it becomes possible to perform backups across the network without causing a major impact on LAN throughput. Filesystems can be remotely mounted over an FDDI LAN with resulting access speeds similar to locally attached disks. File transfer speeds in the range of megabytes per second are also practically achievable on FDDI LANs.
Because FDDI is carried over optical fibres, the opportunities for "tapping" the line are greatly reduced. As light travelling down a fibre does not induce a local magnetic field, as electrical signals do, there is no way to tap the signal through use of an inductive tap. Also, any attempt at splicing a tap into the fibre would cause a disruption in communication which would indicate possible line tampering.
