Biblio5

484 CELLULAR AND PCS RADIO SYSTEMS

16.3.2.2 Okumura Model. Okumura et al. (Ref. 4) carried out a detailed analysis for path predictions around Tokyo for mobile terminals. Hata (Ref. 5) published an empirical formula based on Okumura’s results to predict path loss. The Okumura/ Hata model is probably one of the most widely applied path loss models in the world for cellular application. The formula and its application follow.

A(hr) is the correction factor for mobile antenna height and is computed as follows: For a smallor medium-size city,

where hr is between 1 and 10 m. For a large city,

where ( f ≥ 400 MHz).

Example. Let f c 900 MHz, ht c 40 m, hr c 5 m, and d c 10 km. Calculate A(hr) for a medium-size city.

A(hr) c 12.75 − 3.8 c 8.95 dB

LdB c 69.55 + 72.28 − 22.14 − 8.95 + 34.4 c 145.15 dB.

16.3.2.3 Building Penetration. For a modern multistory office building at 864 MHz and 1728 MHz, transmission loss (LdB) includes a value for clutter loss L(v) and is expressed as follows:

where the attenuation in dB of the floors and walls was af and aw, and the number of floors and walls along the line d were nf and nw, respectively. The values of L(v) at 864 MHz and 1728 MHz were 32 dB and 38 dB, with standard deviations of 3 dB and 4 dB, respectively (Ref. 3).

Another source (Ref. 6) provided the following information: At 1650 MHz the floor loss factor was 14 dB, while the wall losses were 3–4 dB for double plasterboard and 7–9 dB for breeze block or brick. The parameter L(v) was 29 dB. When the propagation frequency was 900 MHz, the first floor factor was 12 dB and L(v) was 23 dB. The higher value for L(v) at 1650 MHz was attributed to a reduced antenna aperture at this frequency compared to 900 MHz. For a 100-dB path loss, the base station and mobile

terminal distance exceeded 70 m on the same floor, was 30 m for the floor above, and 20 m for the floor above that, when the propagation frequency was 1650 MHz. The corresponding distances at 900 MHz were 70 m, 55 m, and 30 m. Results will vary from building to building, depending on the type of construction of the building, the furniture and equipment it houses, and the number and deployment of the people who populate it.

16.4 IMPAIRMENTS: FADING IN THE MOBILE ENVIRONMENT

16.4.1 Introduction

Fading in the mobile situation is quite different from the static line-of-sight (LOS) microwave situation discussed in Section 9.2.4. In this case radio paths are not optimized as in the LOS environment. The mobile terminal may be fixed throughout a telephone or data call, but is more apt to be in motion. Even the hand-held terminal may well have micromotion. When a terminal is in motion, the path characteristics are constantly changing.

Multipath propagation is the rule. Consider the simplified pictorial model in Figure 16.4. Commonly, multiple rays reach the receive antenna, each with its own delay. The constructive and destructive fading can become quite complex. We must deal with both reflection and diffraction.3 Energy will arrive at the receive antenna reflected off sides of buildings, towers, streets, and so on. Energy will also arrive diffracted from knife edges (e.g., building corners) and rounded obstacles (e.g., water tanks, hill tops).

Because the same signal arrives over several paths, each with a different electrical length, the phases of each path will be different, resulting in constructive and destructive amplitude fading. Fades of 20 dB are common, and even 30-dB fades can be expected.

On digital systems, the deleterious effects of multipath fading can be even more severe. Consider a digital bit stream to a mobile terminal with a transmission rate of 1000 bps. Assuming NRZ coding, the bit period would be 1 ms (bit period c 1/ bit rate). We find the typical multipath delay spread may be on the order of 10 ms. Thus

Figure 16.4 Mobile terminal in an urban setting. R c reflection; D c diffraction.

3Diffraction is defined by the IEEE (Ref. 7) as “The deviation of the direction of energy flow of a wave (ray beam), not attributable to reflection or refraction, when it passes an obstacle, a restricted aperture or other inhomogeneities in a medium.

Figure 16.6 Correlation coefficient r versus the parameter h for two receive antennas in different orientations. (From Ref. 8, Figure 6.4, reprinted with permission.)

coefficient but also as a function of antenna height, h. The wider the receive antennas are separated, the lower the correlation coefficient and the more uncorrelated the diversity paths are. Sometimes we find that, by lowering the antennas as well as adjusting the distance between them, we can achieve a very low correlation coefficient. However, we might lose some of the height-gain factor.

Lee (Ref. 8) proposes a new parameter h, where

In Figure 16.6 we relate the correlation coefficient (r) with h, where a is the orientation of the antenna regarding the incoming signal from the mobile unit. Lee recommends a value of r c 0.7. Lower values are unnecessary because of the law of diminishing returns. There is much more fading advantage achieved from r c 1.0 to r c 0.7 than from r c 0.7 to r c 0.1.

Based on r c 0.7 and h c 11, from Figure 16.6 we can calculate antenna separation values (for 850-MHz operation). For example, if h c 50 ft (16 m), we can calculate d using formula 16.4:

d c h/ h c 50/ 11 c 4.5 ft (1.36 m).

For an antenna 120-ft (36.9-m) high, we find that d c 120/ 11 c 10.9 ft (or 3.35 m) (from Ref. 8).

16.4.2.2.1 Space Diversity on a Mobile Platform. Lee (Ref. 8) discusses both vertically separated and horizontally separated antennas on a mobile unit. For the vertical case, 1.5l is recommended for the vertical separation case and 0.5l for the horizontal

16.5.2 Bit Rate Reduction of the Digital Voice Channel

It became obvious to system designers that conversion to digital cellular required some different technique for coding speech other than conventional PCM, found in the PSTN and described in Chapter 6. The following lists some techniques that have been considered or that have been incorporated in the various systems in North America, Europe, and Japan (Ref. 11):

1. ADPCM (adaptive differential PCM). Good intelligibility and good quality; 32kbps data transmission over the channel may be questionable;

2. Linear predictive vocoders (voice coders); 2400 bps. Adopted by U.S. Department of Defense. Good intelligibility, poor quality, especially speaker recognition;

3. Subband coding (SBC). Good intelligibility, even down to 4800 bps. Quality suffers below 9600 bps;

4. RELP (residual excited linear predictive) type coder. Good intelligibility down to 4800 bps and fair to good quality. Quality improves as bit rate increases. Good quality at 16 kbps;

5. CELP (codebook-excited linear predictive). Good intelligibility and surprisingly good quality, even down to 4800 bps. At 8 kbps, near-toll quality speech.

16.6 NETWORK ACCESS TECHNIQUES

16.6.1 Introduction

The objective of a cellular radio operation is to provide a service where mobile subscribers can communicate with any subscriber in the PSTN, where any subscriber in the PSTN can communicate with any mobile subscriber, and where mobile subscribers can communicate among themselves via the cellular radio system. In all cases the service is full duplex.

A cellular service company is allotted a radio bandwidth segment to provide this service. Ideally, for full-duplex service, a portion of the bandwidth is assigned for transmission from a cell site to mobile subscriber, and another portion is assigned for transmission from a mobile user to a cell site. Our goal here is to select an “access” method to provide this service given our bandwidth constraints.

We will discuss three generic methods of access: (1) FDMA, (2) TDMA (time division multiple access), and (3) CDMA (code division multiple access). It might be useful for the reader to review our discussion of satellite access in Section 9.3, where we described FMDA and TDMA. However, in this section, the concepts are the same, but some of our constraints and operating parameters are different. It also should be kept in mind that the access technique has an impact on overall cellular bandwidth constraints. TDMA and CDMA are much more efficient, achieving a considerably greater number of users per unit of RF bandwidth than FDMA.

16.6.2 Frequency Division Multiple Access (FDMA)

With FDMA our band of RF frequencies is divided into segments and each segment is available for one user access. Half the contiguous segments are assigned to the cell site for outbound traffic (i.e., to mobile users) and the other half to inbound. A guardband

Figure 16.9 A TDMA delay scenario.

Suppose there are ten users. Let each user’s bit rate be R, then a user’s burst must be at least 10R. Of course, the burst will be greater than 10R to accommodate a certain amount of overhead bits, as shown in Figure 16.8.

We define downlink as outbound, base station to mobile station(s), and uplink as mobile station to base station. Typical frame periods are:

One problem with TDMA, often not appreciated by many, is delay. In particular, this is delay on the uplink. Consider Figure 16.9, where we set up a scenario. A base station receives mobile time slots in a circular pattern and the radius of the circle of responsibility of that base station is 10 km. Let the velocity of a radio wave be 3 × 108 m/ s. The time for the wave to traverse 1 km is 1000 m/ (3 × 108) or 3.333 ms. In the uplink frame we have a mobile station right on top of the base station with essentially no delay and another mobile right at 10 km with 10 × 3.33 ms or 33.3 ms delay. A GSM time slot is about 576 ms in duration. The terminal at the 10-km range will have its time slot arriving 33.3 ms late compared to the terminal with no delay. A GSM bit period is about 3.69 ms so that the late arrival mutilates about 10 bits and, unless something is done, the last bit of the burst will overlap the next burst (Refs. 3, 12).

Refer now to Figure 16.10, which illustrates GSM burst structures. Note that the access burst has a guard period of 68.25 bit durations or a slop of 3.69 × 68.25 ms, which will well accommodate the later arrival of the 10-km mobile terminal of only 33.3 ms.

To provide the same long guard period in the other bursts is a waste of valuable “spectrum.”6 The GSM system overcomes this problem by using adaptive frame alignment. When the base station detects a 41-bit random access synchronization sequence with a long guard period, it measures the received signal delay relative to the expected signal from a mobile station with zero range. This delay, called the timing advance, is transmitted to the mobile station using a 6-bit number. As a result, the mobile station advances its time base over the range of 0 –63 bits (i.e., in units of 3.69 ms). By this process the TDMA bursts arrive at the base station in their correct time slots and do

6We are equating bit rate or bit durations to bandwidth. One could assume 1 bit/ Hz as a first-order estimate.

492 CELLULAR AND PCS RADIO SYSTEMS

Figure 16.10 GSM frame and burst structures. (From Ref. 3, Figure 8.7. Reprinted with permission.)

not overlap with adjacent ones. As a result, the guard period in all other bursts can be reduced to 8.25 × 3.69 ms or approximately 30.46 ms, the equivalent of 8.25 bits only. Under normal operations, the base station continuously monitors the signal delay from the mobile station and thus instructs the mobile station to update its time advance parameter. In very large traffic cells there is an option to actively utilize every second time slot only to cope with the larger propagation delays. This is spectrally inefficient but, in large, low-traffic rural cells, admissible (from Ref. 3).

16.6.3.1 Comments on TDMA Efficiency. Multichannel FDMA can operate with a base station power amplifier for every channel, or with a common wideband amplifier for all channels. With the latter, we are setting up a typical generator of intermodulation (IM) products as these carriers mix in a comparatively nonlinear common power amplifier. To reduce the level of IM products, just like in satellite communications discussed in Chapter 9, backoff of the power amplifier is required. This backoff can be in the order of 3–6 dB.

With TDMA (downlink), only one carrier is present on the power amplifier, thus removing most of the causes of IM noise generation. Thus with TDMA, the power amplifier can be operated to full saturation, a distinct advantage. FDMA required some guardband between frequency segments; there are no guardbands with TDMA. However, as we saw previously, a guard time between uplink time slots is required to accommodate the following situations:

•Timing inaccuracies due to clock instabilities;

•Delay spread due to propagation;7

7Delay spread is a variance in delay due to dispersion of emitted signals on delayed paths due to reflection, diffraction/ refraction. Lee reports a typical urban delay spread of about 3 ms.

•Transmission delay due to propagation distance (Section 16.6.3); and

•Tails of pulsed signals due to transient response.

The longer guard times are extended, the more inefficient a TDMA system becomes.

16.6.3.2 Advantages of TDMA. The introduction of TDMA results in a much improved transmission system and reduced cost compared to an FDMA counterpart. Assuming a 25-MHZ bandwidth, up to 23.6 times capacity can be achieved with North American TDMA compared to FDMA, typically AMPS (see Ref. 13, Table II.)

A mobile station can exchange system control signals with the base station without interruption of speech (or data) transmission. This facilitates the introduction of new network and user services. The mobile station can also check the signal level from nearby cells by momentarily switching to a new time slot and radio channel. This enables the mobile station to assist with handover operations and thereby improve the continuity of service in response to motion or signal fading conditions. The availability of signal strength information at both the base and mobile stations, together with suitable algorithms in the station controllers, allows further spectrum efficiency through the use of dynamic channel assignment and power control.

The cost of base stations using TDMA can be reduced if radio equipment is shared by several traffic channels. A reduced number of transceivers leads to a reduction of multiplexer complexity. Outside the major metropolitan areas, the required traffic capacity for a base station may, in many cases, be served by one or two transceivers. The saving in the number of transceivers results in a significantly reduced overall cost.

A further advantage of TDMA is increased system flexibility. Different voice and nonvoice services may be assigned a number of time slots appropriate to the service. For example, as more efficient speech CODECs are perfected, increased capacity may be achieved by the assignment of a reduced number of time slots for voice traffic. TDMA also facilitates the introduction of digital data and signaling services as well as the possible later introduction of such further capacity improvements as digital speech interpolation (DSI).

16.6.4 Code Division Multiple Access (CDMA)

CDMA means code division multiple access, which is a form of spread spectrum using direct sequence spreading (see Ref. 14). There is a second class of spread spectrum called frequency hop, which is used in the GSM system, but is not an access technique.

Using spread spectrum techniques accomplishes just the opposite of what we were trying to accomplish in Section 9.2.3.5. There bit packing was used to conserve bandwidth by packing as many bits as possible in 1 Hz of bandwidth. With spread spectrum we do the reverse by spreading the information signal over a very wide bandwidth.

Conventional AM requires about twice the bandwidth of the audio information signal with its two sidebands of information (i.e., approximately ±4 kHz).8 On the other hand, depending on its modulation index, frequency modulation could be considered a type of spread spectrum in that it produces a much wider bandwidth than its transmitted information requires. As with all other spread spectrum systems, a signal-to-noise advantage is gained with FM, depending on its modulation index. For example, with AMPS, a typical FM system, 30 kHz is required to transmit the nominal 4-kHz voice channel.

8AM for “toll-quality” telephony.