1.5 Inuence of the Channel on Space-Time Pro- cessing
Space-time processing algorithms are profoundly inuenced by the channel characteristics6. A description of the eects of the channel characteristics and the corresponding mitigation techniques are given in Figure 5.6.
-
- Reduces reciprocity in TDD - Time varying channel - Time selective fading
- ISI
- Reduces reciprocity in time - Frequency selective fading
Space selective fading channel
Mitigation
- Reduces reciprocity in space channel
- Channel tracking - Reduce TDD turn around - Time diversity
time
- Equalization/RAKE - Frequency diversity - Angle selectivity
- Space diversity
- Reduce frequency spread in FDD
Angle Spread
Effect
Doppler Spread
Delay Spread
Figure 1.9: Channel characteristics inuencing space-time processing.
6In space-time processing, the channel is broadly dened to include also the interference channels.
1.5.1 Doppler Spread
Doppler spread, induced by the motion of subscribers or scatterers, has a strong inuence on space-time processing algorithms in dierent dimen- sions. The Doppler spread is large in macro-cells which serve high mobility subscribers. Also it increases with higher operating frequencies. Doppler spread is also present in low mobility (microcell) or xed wireless networks due to mobility of scatterers (e.g. trac). A discussion of Doppler and delay spreads of mobile radio channels can be found in [51].
In a TDMA system, if the time period of a frameis small compared to the coherence time of the channel (as in GSM), the channel will be reasonably constant during the frame, and we do not need to track the channel during the frame. On the other hand if the frame duration is comparable to, or longer, than the coherence time of the channel (as in IS-136), the channel changes signicantly and we need to track the channel during the frame.
Although the channel typically does not vary very much during a frame in GSM it can vary a non-negligible amount if the speed is very high, say on a high speed train, and the carrier frequence is high, say 1800 MHz. In this case some improvements can be achieved by designing an equalizer which will be robust against the anticipated time variations during the frame. How to design such equalizers is discussed in Section 7.3 of Chapter 7.
Fading can sometimes be combatted in the time domain by interleaving and coding. This is however only eective if the coherence time is shorter than the interleaver depth. For slowly time varying channels, other forms of di- versity may be necessary to ensure acceptable link quality, e.g. frequency hopping. Also, as mentioned in Section 1.4.1, in time division duplex sys- tems, the reciprocity of the channel is valid only if the channel coherence time is much larger than the duplexing time.
1.5.2 Delay Spread
Delay spread arises from multipath propagation and can be large in macro- cell systems with antennas located above the roof tops. It is most pro- nounced in hilly terrain areas and least pronounced in at rural terrain installations. Microcells using antennas mounted below the roof tops, tend
1.5. Inuence of the Channel on Space-Time Processing 23 to have small delay spreads.
Delay spread aects space-time processing algorithms in several ways. In TDMA systems, if the symbol period is much shorter than the delay spread of the channel, we can avoid equalizers (as in PACS and PHP). In contrast, in GSM, the delay spread can be much larger than the symbol period, man- dating the use of equalizers. In general, combined space and time processing is more eective for delay spread mitigation than time processing alone.
Likewise, in CDMA, if the delay spread is larger than the chip period, we have inter-chip interference which, however, is usually less insidious than the intersymbol interference in TDMA. Typically, the diversity in paths is exploited by a RAKE receiver.
Delay spread in the channel will increase the number of uncorrelated signals impinging on an antenna array. If there is no delay spread in the channel, then the number of uncorrelated signals will be equal to the number of users, desired and undesired, transmitting to the antenna array. Note however the transmitted signals are assumed to be temporally white. Thus, if the channels from the dierent users to the antenna array has a delay spread, then the number of uncorrelated signals will be increased. Roughly we can say that each delay spread of a symbol interval adds one uncorrelated signal per user.
1.5.3 Angle Spread
Angle spread arises from multipath arrivals from dierent directions. It is largest at the subscriber unit, where local scatterers may result in 360 degrees angle spread. At the base station, the angle spread is large in mi- crocells with below roof top antennas. Base stations in macro-cells witness less angle spread, it is the lowest in rural regions and becomes signicant in urban and hilly regions.
Angle spread inuences a number of space-time processing issues. First, high angle spread increases spatial diversity which should be exploited by space-time processing. Next, the reciprocity of the spatial signature of the channel is reduced if the angle spread is large [84]. The angle spread also reduces the eectiveness of methods parametrizing the signal in directions
of arrival since these will be less distinct.
The angle spread and the delay spread will aect the spatio-temporal struc- ture of the channel. The spatio-temporal structure of the channel can be either coupled or can decouple. If we only have angular spread and no delay spread in the propagation channel then the overall channel will have a de- coupled spatio-temporal structure. The channel can then be described by a temporal lter, representing the temporal pulse shaping in the transmitter and the receiver followed by a purely spatial SIMO lter representing the spatial spreading of the propagation channel. The spatio-temporal structure of the channel aects the structure and complexity of appropriate space-time equalizer. This is discussed more in detail in Section 5.3 of Chapter 5.
1.5.4 Dierent Realizations of a Multi-Channel Receiver
A multi-channel receiver can be realized by using multiple antenna elements that are spatially separated. There is however many other ways of realizing a multi-channel receiver.
Polarization diversity is an obvious way to realize multiple channels. Al- though the two polarizations of the signal may be collected from the same spatial location, they typically have encountered independently fading chan- nels. In most of the methods we use here we can simply view the signals as coming from two independent antenna elements.
Fractionally spaced sampling can be of interest if the symbol sampling fre- quence is lower than twice the maximum frequence content of the signal.
Fractionally spaced sampling can be handled as a multi-channel receive prob- lem. The multiple samples during a symbol interval can be treated as if they were coming from dierent sensors. The channels will be correlated, but not identical and can thus aid the equalization or the symbol detection. This can be viewed as an analogy to the multiple antenna receiver and many of the techniques applied to spatio-temporal processing can be utilized in together with fractionally spaced sampling.
Partitioning into real and imaginary part: When the modulation format only utilizes one of the dimensions in the complex plane, for example if we have binary phase shift keying modulation, then the real and the imaginary part