Capacity in a Cellular System
Capacity in a Cellular System
Capacity is a concern in any wireless communications system. High demand for cellular service, especially in large urban markets, has created a need to serve a greater number of users in a limited amount of frequency space. Cellular system operators are looking for new ways to fit more users into their increasingly crowded network, and many are choosing to move from the existing analog transmission technology to one of the competing digital standards. These standards are also being selected by the new PCS providers as they begin to build out their own networks. Although digital systems provide a variety of benefits, this month we'll focus on two main digital access methods and their effect on system capacity.
Almost all current and proposed digital standards are based on either Time Division Multiple Access (TDMA) or Code Division Multiple Access (CDMA). These two methods are fundamentally different and incompatible with each other, but each claim to able to support anywhere from three to more than twenty times the number of simultaneous callers than the current AMPS cellular standard, which uses Frequency Division Multiple Access (FDMA).
Frequency Division Multiple Access
FDMA is a familiar method of allocating bandwidth. A band of frequencies is divided up in to channels, each of a particular size. A transmitter in an FDMA system is given exclusive use of one or more channels. This is how broadcast radio and television are set up - each "owns" a portion of the frequency spectrum that is set aside just for them. This is also true of cellular systems, where a base station is allowed to transmit on a set of forward channels (see the November column for details) and a mobile unit transmits on one of a set of reverse channels. No other base station within range of the mobile will be transmitting on the same forward channel, and no other mobile within range of the base station should be transmitting on the same reverse channel. Both the base and the mobile usually transmit continuously during a conversation, and fully occupy their assigned forward and reverse channels. No other conversation can take place on these channels until the first conversation is completed (or the call is handed off to another base station).
Time Division Multiple Access
TDMA is a more efficient, but more complicated way of using FDMA channels. In a TDMA system each channel is split up into time segments, and a transmitter is given exclusive use of one or more channels only during a particular time period. For example, in North American TDMA (also known as Interim Standard 54) each channel is essentially divided into three timeslots. A maximum of three transmitters take turns sending in their assigned timeslots. A conversation, then, takes place during the time slots to which each transmitter (base and mobile) is assigned. TDMA requires a master time reference to synchronize all transmitters and receivers.
TDMA was the first digital standard to be proposed, and is attractive to current cellular operators because it allows existing analog customers to continue using the network as digital capability is added. Under IS-54, channel bandwidth remains 30 kHz and the data format on the control channels remains identical to those under AMPS. Voice channels are designated as either analog or digital, and analog cell phones continue to operate as they do under AMPS since they will always be assigned to analog voice channels. Dual-mode and digital-only phones are assigned digital voice channels where they are available. Base stations can be converted to digital capability as demand and funding allow.
Code Division Multiple Access
CDMA is fundamentally different than TDMA and FDMA. Where FDMA and TDMA transmit a strong signal in a narrow frequency band, CDMA transmits a relatively weak signal across a wide frequency band. Using a technique called direct sequence spread spectrum, the data to be transmitted are combined with a pseudo-noise code (a pre-determined binary sequence that appears random) and transmitted broadband. CDMA under Interim Standard 95 uses a bandwidth of 1.25 MHz, or more than 40 times the 30 kHz channels of AMPS and TDMA.
The pseudo-noise code (PN code) is a series of binary "chips" that are much shorter in duration than the data bits. Since the chips appear to be in a random pattern, and there are many chips per data bit (in IS-95 there are 128 chips for each data bit), the modulated result appears to normal (FDMA) receivers as background noise. A spread spectrum receiver using the same PN code as the transmitter will correlate, or match up, the wide band signal with the chip sequence and despread the signal, recovering the data. A spread spectrum receiver with a different PN code will not be able to recover that signal, and if the PN codes were chosen correctly, will hear nothing but noise. This relative immunity to interference, whether from outside sources or other spread spectrum transmitters, gives CDMA systems the ability to pack many users into the same frequency space at the same time. It also gives a measure of security to each signal, since each user will have a different PN code. CDMA also does not require different base station radios for each user - the same radio may serve multiple users with just a change in PN code.
There is a limit to CDMA capacity, however, and it is essentially the amount of interference a CDMA receiver can tolerate. As more and more units transmit, the amount of noise a receiver sees goes up, since all signals not using the receiver's specific PN code appear as noise. At some point there is so much noise that the receiver can no longer hear the transmitter. Boosting the transmitter power won't help overall, since it increases the noise for all the other receivers, who would in turn tell their transmitters to boost power, and the situtation remains.
A similar real-world issue CDMA has to face is known as the near-far problem. In a nutshell, if a unit near a base station is transmitting with too much power, signals from units far from the base station will be lost in the noise. CDMA system designers believe they have this problem solved by being able to rapidly adjust the amount of power each unit is using to transmit. This already occurs to some degree in AMPS systems, where the base station commands far away units to use more power and near units to use less power. This process must occur very quickly in CDMA systems, and it remains to be seen whether it can be made to work reliably outside of the laboratory.
And now, an Analogy
It may be easier to visualize the differences between all these methods by imagining a cocktail party where two people wish to converse with each other. With FDMA, everyone in the room must be silent except for the speaker. The speaker may talk as long as they wish, and when they finish someone else may start speaking, but again only one at a time. New speakers must wait (or find another party) for the current speaker to finish before starting. Everyone in the room can hear and understand the speaker, unless they are too far away or the speaker's voice is too soft. If the intended listener is close enough, the speaker may decide to whisper. Conversely, if the listener is too far away, the speaker may have to shout. Since no one else should be talking, this presents no problem. If someone talks out of turn, the listener will probably be confused and not be able to understand either speaker.
In TDMA, everyone in the room agrees to watch a clock on the wall, and speak only during a particular time. Each person wishing to talk is given a set period of time, and each person listening must know what that time period will be. For example, everyone may agree on time slots with a duration of ten seconds. Speaker number one may talk for ten seconds starting from the top of the minute. The listener who wishes to hear this speaker must also be made aware of the schedule, and be ready to listen at the top of the minute. Speaker number two may speak only from ten seconds after the minute until twenty seconds after. As with FDMA, only one person at a time may speak, but each speaker's time is now limited and many persons may take their turn. If someone in the room cannot see the clock, they will not be able to speak and will have great difficulty understanding the speakers.
In CDMA, the speaker and the listener have agreed beforehand to use a language that no one else at the party understands. Many speakers may talk at a CDMA party, each using a different language, and it is relatively easy for the listener to hear and understand the speaker as long as there aren't too many speakers talking at the same time. As more and more speakers start talking, the noise level in the room goes up and it becomes harder and harder for the listener to make out what their speaker is saying. If a speaker begins to shout, in order for their listener to hear better, it raises the noise level even more and makes it more difficult for everyone else in the room.
Which is Better?
The question of which digital access method is better has been debated for years, and will continue to be the source of much disagreement. It is often difficult to sort fact from opinion, since many sources of information come from manufacturers who have a vested financial interest in the success of one standard over another. TDMA systems, of which GSM is one, are in place and operating and have been successful for many years. CDMA systems have provided the military with secure, jam-resistant communications for decades, but have yet to prove themselves capable of delivering the promised capacity increases in a commercial environment. Both sides claim to provide better voice quality and lower cost of service than the other. As with many questions involving the fast-paced world of communications, keeping informed and knowing the underlying principles will help sort things out.
Speaking of keeping informed, if you have access to the World Wide Web, be sure to check out the _PCS Frontline_ web pages at http://www.decodesystems.com. You'll find information about cellular and PCS systems, including frequency allocations, channel assignments, and details that didn't make it into this column. As always, electronic mail is welcome at firstname.lastname@example.org. Until next month, happy monitoring!
Click here for the index page.
Click here for the main page.