The types of systems discussed thus far have implicitly assumed circuit switching. That is, when a user starts to make a call, a circuit is established between the user and the network, which is maintained for the duration of the call. The circuit may be an FDMA channel, a timeslot on a TDMA channel, or a CDMA orthogonal code. Whatever, the net effect is that nobody else is able to use that particular resource for the duration of the call.
An alternative to circuit switching is packet switching. In a packet switched system, no permanent connection is established. Instead, the subscriber unit collects data from the user until its buffer is full, then it requests a short slot from the network to transmit the packet of data. The unit then relinquishes the network resources and waits for the buffer to fill again. Packet switching comes in two guises, connection-oriented and nonconnection-oriented. In the case of connection-oriented packet switching, a virtual circuit is established between the transmitter and the receiver, passing through the switching nodes, when the first packet is received. All subsequent packets received for the same destination travel via the same route. Further, they are received in the order in which they are transmitted. In the case of nonconnection-oriented packet switching, each packet is treated as if no previous packet had been sent. Potentially, a packet could be sent via a different route from the previous packet and the packets might not arrive at the receiver in the order they were sent.
The receiver then requires a sufficient buffer so it can correctly order the data prior to presenting it to the user.
In outline, circuit switching provides a low and known delay but uses resources inefficiently compared to packet switching. Broadly speaking,
circuit switching is suitable for voice, while packet switching is suitable for data. Packet switching is unsuitable for voice because the delays suffered by each packet can be variable, resulting in significant and unwanted voice delay. Packet switching is particularly suitable when the data to be transmitted arrives in short bursts and short delays can be tolerated. That is illustrated by the following example.
Imagine a data source that provided data at the rate of 200 bits every 3 sec and required that the delay on transmission was less than 2 sec (e.g., vehicle location systems provide data in this manner). If the data was transmitted via a circuit switched channel with, say, a data capacity of 9.6 Kbps, a call setup time of 1.5 sec, and a call clear-down time of 1 sec, it would be necessary to maintain the data channel dedicated to that use.
If it was attempted to clear down the call between bursts, the signaling required to do so would take so long that the subscriber unit would need to reestablish immediately the channel once it had been cleared down.
However, using a packet protocol, with an overhead of 20%, only 200/3× 120% bps=80 bps would be transmitted. That requires less than 1% of the available channel capacity. Packet-mode systems also are ideal for asymmetrical applications where more data is transmitted in one direction than the other (e.g., Internet browsing). Because the uplink and downlink need not be paired, uplink resources are freed for another user who may want to send a large data file in the uplink direction but receive little in the downlink direction.
The ideal radio system probably would include both circuit and packet switched capabilities. Indeed, some modern mobile radio systems, including GSM, are being developed with such dual capabilities. At the time of this writing, all the available WLL systems are circuit switched, but some manufacturers are considering the development of a packet switched capability.
References
[1] Steele, R., ed.,Mobile Radio Communications, New York: Wiley & Sons, 1992.
[2] Webb, W., and L. Hanzo,Modern Quadrature Amplitude Modulation, New York:
Wiley & Sons, 1994.
118 Introduction to Wireless Local Loop
8
TDMA or CDMA?
S , there has been substantial debate in the mobile radio journals and more recently in publications addressing WLL about whether CDMA or TDMA is the best access technique. The debate has tended to be distorted and acrimonious because the protago- nists are manufacturers of the different access technologies, who want to prove to operators that their equipment is superior. This chapter intro- duces some of the key aspects to the debate, focusing specifically on WLL.
It should be noted that after six years and the application of most of the finest minds in mobile radio, the debate still has not been resolved.
It would be optimistic to expect this chapter to provide all the answers.
What it can do is put the discussion on an independent and rational footing and note the key issues and implications for WLL.
The debate has focused mostly on whether CDMA can provide greater capacity for a given spectrum allocation than TDMA. That is clearly a key issue for a WLL operator and is explored in detail in Section 8.1. Other parts of the debate have looked at factors such as ease
119
of cell planning and signal quality and flexibility; each of those factors is addressed in Section 8.2. Finally, Section 8.3 summarizes the implica- tions for WLL, which will be picked up when the separate technologies are examined in Part IV.