In addition, we suppose an uncoded transmission for TDMA systems and coded transmission for DS-CDMA systems

( rc = 1 / 2 ). The reason for assuming a convolutional encoding for the case of DS-CDMA systems is that the benefit of FEC are obtained in these systems without any associated reduction in capacity. the assumed bit energy to noise spectral

density ratios Eb / N0 for a given BER are reported in table 5.

6 TDMA systems

6.1Broadband TDMA

Broadband TDMA systems generally permit the time sharing of the same frequency band by different users.

The physical layer is normally based on 2 Mbit/s transmission, with a total capacity equivalent to a 2 Mbit/s PCM system but additional signalling to control the radio link. Radio carrier spacing is typically 1,75 MHz or 2 MHz. Systems use frequency-division-duplex, and require an appropriate paired frequency allocation. Some systems are also available using 4 Mbit/s or 8 Mbit/s transmission and providing 60 or 120 traffic channels, with 3,5 MHz or 7 MHz channel spacing.

Access is by TDM (downlink), TDMA (uplink), demand assigned, with frequency FDD. The CS transmits, continuously, a frame structure which is based on PCM but with normal PCM slots concatenated to longer radio slots: the downlink radio frame length is typically 1 to 10 ms (depending on system), much longer than the normal 125 s PCM frame. The uplink is burst-mode: the reason for concatenating PCM slots is to make the active up-slot length longer and reduce the effect on efficiency of the required up-link guard periods. Adjustable timing advance (e.g. GSM) is provided to make sure all incoming bursts at the central station fit exactly in their allocated slots.

ETSI

PCM frame stream

Inbound 64 kbit/s

stream

5 ms

Figure 4: TDMA frame format

Figure 4 shows diagrammatically a possible frame format for a point-to-multipoint system, and how a single 64 kbit/s user accesses the system. On the down-link, a 2,048 Mbit/s PCM stream, which is augmented with additional control information for radio access, is mapped onto the down frame structure. The down-link frame has 32 slots, but in this example its length is 5 ms instead of the normal PCM frame of 125 s. Therefore the contents of time slot 31 (for example) of 40 successive PCM frames are stored and loaded into time slot 31 of the TDMA frame. Similarly, on the up-link, a single 64 kbit/s tributary stream will be stored for each 5 ms frame period and loaded into a single, 156,25 s up-link burst. The timing of this burst will be adjusted so that it is received at the central station in synchronism with the down-link frame.

The radio architecture assumes that TS radio transceivers have to be full duplex, and therefore it is not necessary for the downlink and uplink frame lengths to be equal; this may enable a better trade-off of signal delay against minimum overhead in the frame structure.

Since the protocol is based on PCM, traffic is carried in 64 kbit/s channels, and the system is therefore fully transparent to normal speech-band services. Certain suppliers provide terminal equipment with full ISDN 2B+D capability. Terminal equipment is capable of supporting either single subscribers, or groups of subscribers, is generally available.

Systems operate typically at 1,5 GHz or 2,4 GHz in normal "fixed-"reuse" frequency allocations, and frequency rasters are aligned with microwave P-P systems.

Central station to subscriber system range can be typically up to 40 km. Both are installed to obtain a good line-of-sight path. In addition, most systems make provision for a "repeater" mode which allows ranges of several hundred, kilometres to be attained in several "hops".

For systems specifically designed for data applications, packet data transmission can be deployed using contention avoidance protocols.

6.2Narrowband TDMA

A typical 2 Mbit/s P-MP system as described above according to EN 301 021 [7] has a threshold receive sensitivity of -90 dBm and peak transmit power of +35 dBm (though the transmit power would probably not exceed +30 dBm in practice). The total link budget, assuming CS antenna gain of 10 dBi and TS antenna gain of 18 dBi, would be:

30 + 10 + 18 - (-90) = 148 dB.

ETSI

At 2 Mbit/s the symbol rate will be 1 Mbaud with a quaternary modulation scheme.

Comparing these figures with the propagation measurements for ad-hoc TS deployment reported in annex A, clearly quite a high proportion of the surveyed points would not receive service because of excessive path loss. Moreover, some measured path impulse responses are significantly longer than the symbol length for a 2 Mbit/s system, so that a high error rate due to intersymbol interference caused by multipath propagation will arise for some TS locations.

For optimum operation with "ad-hoc" deployment, a system with a lower gross RF channel bit rate will be preferable. This allows a better link budget, through a smaller channel bandwidth; makes the system less susceptible to multipath distortion because the symbol rate is lower; and makes it feasible and economical to implement an adaptive equalizer to correct any multipath.

For example, with a minimum channel bit-rate of 0,32 Mbit/s, the system noise bandwidth is reduced by a factor of 6,25, giving 8 dB increase in the link budget relative to a "2 Mbit/s" system; the symbol length (assuming quaternary modulation) is 4 to 6 times longer (depending on the amount of transmission overhead). The example of GSM, which has approximately the same symbol length, shows that a low-cost equalizer realization is feasible for such a symbol rate.

The use of a narrow-band channel leads to a lower system capacity for a single-carrier system. At a minimum bit-rate of 0,32 Mbit/s, five 64 kbit/s traffic channels may be accommodated per carrier. However, the lower bit rate also allows the channel spacing to be significantly reduced. Using frequency stabilization techniques in the CS and TS whereby the transmit and receive channel frequencies are locked to the bit-rate, channel spacing as low as 300 kHz may be employed.

If larger system capacities are then required (e.g. if the CS has to support an El link capacity), multiple radio carriers can be radiated. Clearly 6 carriers would be required to support 30 × 64 kbit/s traffic channels, to be equivalent to a conventional 2 Mbit/s P-MP system. On the other hand, it may be useful for the operator to conserve spectrum by installing fewer carriers to reduce system capacity where penetration is low or where dictated by the constraints of cellular frequency planning.

The total occupied bandwidth for the narrow-band system would be 1,8 MHz, compared to 1,75 MHz or 2,0 MHz for the conventional approach - that is, the systems will be very similar in terms of occupied bandwidth. Greater effective spectral efficiency may also be obtained by using lower rate coding.

It should be noted that, for a given multipath delay spread the processing power of an adaptive equalizer, measured in for example MIPS, has to rise as the square of the system bit rate. This is because the sampling rate varies proportionally to the bit rate, and the number of equalizer taps (assuming a transversal equalizer) also goes up proportionally to the bit rate for a length of impulse response.

For speech traffic this is appropriate, as long as an effective service transparency can be provided, for both narrow-band and broad-band systems.

6.3Isolated performance for broadband TDMA

As stated in previous paragraph, physical layer for a broadband TDMA systems is normally based on 2,048 Mbit/s transmission, with a total capacity equivalent to 32 traffic channels. Some systems are also available using 4,096 Mbit/s

ETSI

Substituting these values the following characteristics of TDMA system in absence of interfering cells result as in table 6.

Table 6: Broadband TDMA capacity (number of 64 kbit/s channels) and threshold power level without interfering cells

For system supporting packed data transmission for data applications, FSK method should be considered. Table 7 shows the bit energy to noise spectral density ratios Eb/N0 for various FSK modulation schemes with incoherent demodulation.

Table 7: Assumed Eb/N0 for FSK modulation schemes versus BER

The corresponding sensitivities are shown in table 8.

Table 8: Broadband TDMA capacity and threshold power level without interfering cells for systems using incoherent FSK modulation schemes

ETSI