GB2349549A

GB2349549A - Adaptive delay spread window for receiver operating in a multipath environment

Info

Publication number: GB2349549A
Application number: GB9909766A
Authority: GB
Inventors: Andrew Robert Bell; John Vernon Anderton
Original assignee: NEC Technologies UK Ltd
Current assignee: NEC Technologies UK Ltd
Priority date: 1999-04-29
Filing date: 1999-04-29
Publication date: 2000-11-01
Anticipated expiration: 2019-04-29
Also published as: GB2349549B; GB9909766D0

Abstract

In a digital mobile communication system in which the channel between a base station and a mobile is subject to varying degrees of multipath fading, a channel estimator in the receiver at the mobile makes measurements over a delay spread window set at a value long enough to take into account the delay spread in the time between the first signal to arrive at the receiver and the last significant delayed version of the signal to arrive at the receiver. By adaptively varying the delay spread window, rather than setting it at a fixed worst case value, the complexity of the channel estimation process may be reduced. The delay spread window is initially set in accordance with an estimate of the shortest path length between the base station and the mobile (steps 1/1, 1/2). The current actual delay spread is then measured by the channel estimator (step 1/3) and the shortest path length is re-estimated (step 1/4). The delay spread window is then reset (step 1/5) according to the measured actual delay spread and reestimated shortest path length. Steps 1/3 to 1/5 are repeated so that the delay spread window size is continuously adapted. Decisions to switch from one fixed window size to another may be implemented at threshold estimates of the shortest path length. In a CDMA system (such as UMTS), signal path length may be estimated from path loss which is determined according to power loss between the actual power transmitted by the base station (indicated to the mobile by a word in the control channel) and the received power at the mobile. In a TDMA system (such as GSM), path length may be estimated from the timing advance.

Description

Delay Estimation in digital radio systems.

This invention relates to digital mobile telecommunications and in particular it relates to delay spread estimation for telecommunications stations operating in a multipath environment.

The following abbreviations are used herein: GSM-Global Systemfor Mobile communications, (formerly Groupe Special Mobile) TDMA-Time division multiple access CDMA-Code division multiple access UMTS-Universal Mobile Telecommunications System BS-Base station MS-Mobile station FDD-Frequency division duplex In mobile communications the physical channel between base station and mobile station is subject to varying degrees of multipath fading. The effect of multipath is that a number of versions of the transmitted signal arrive at the receiver at different times, each version having taken a different path from transmitter to receiver. The time interval between the first signal to arrive at the receiver and the last significant delayed version of the signal to arrive at the receiver is known as delay spread.

The first signal to arrive at the receiver may have travelled along the line of sight from transmitter to receiver and in this circumstance the signal path length will be the same as the distance between transmitter and receiver.

Where a line of sight signal path does not exist between transmitter and receiver, however, the first signal to arrive at the receiver may have a signal path length much longer than the line of sight distance. As the path loss estimates referred to herein are based upon the reduction of power along the signal path then the term"distance"equates herein to signal path length rather than the line of site distance between MS and BS.

Different types of multiple access schemes (e. g. TDMA for GSM, CDMA for UMTS, etc.) use different methods of channel estimation and compensation but in all of the different methods the complexity of the estimation process increases in relation to the increase in the delay spread over which the measurement is performed.

Technological background for delay spread detection, equalisation and compensation in digital communications is discussed in e. g. US 5 550 868 to Boccuzzi, US 5 479 446 to Mourot, US 5 375 143 to Kazecki et al, US 5 659 576 to Critchlow et al and W098/02967 (Lilleberg et al).

It is an object of the invention to simplify the delay estimation process by varying the period of time over which measurements are taken in accordance with estimates of current distance from transmitter to receiver and modified by currently measured delay spread. The delay spread length adopted for any particular multipath signal input to the receiver will be referred to herein as the delay spread window.

According to the invention there is provided a delay spread estimation device for a receiver in a digital radio system in which the size of a delay spread window and start time of search is set in accordance with an estimate of the shortest signal path length from transmitter to receiver computed within the receiver and modified in accordance with current measures of delay spread and re-estimates of shortest signal path length.

Examples of the invention will now be given with reference to the figures in which: figure 1 is a flowchart illustrating the implementation of the invention in a CDMA system such as UMTS, figure 2 is a flowchart illustrating the implementation of the invention in a TDMA system such as GSM, figure 3 is a graph showing estimates of path loss versus shortest signal path.

The multipath delay time dispersion experienced by the MS will depend on the set of signal paths from BS to MS at any instant. There is a variation in the delay time dispersion which depends on the particular set of signal paths existing at any time between the MS and BS. In order to accommodate the delay time dispersion for all situations, however, most systems have a delay time dispersion window set to a"worst case"value e. g. +/-16 microseconds for GSM. The worst case value should therefore take into account the various environments in which the MS is expected to operate so as to obtain all of the data representing the transmission channel.

The complexity associated with the channel estimation process increases as the size of the search window increases. There appears to exist, however, a relationship between the worst case delay spread at the MS and the current distance of the MS from the BS. An increase of delay spread may be assumed generally to occur as the distance between MS and BS increases.

As an estimate of the signal path length from the BS to the MS may be computed in the receiver, then an estimate of the most likely current worst case delay spread based on signal path length will allow adoption of the most suitable initial delay spread window. This will allow, in general, a smaller size search window corresponding to the current most likely worst case delay spread thereby providing for a reduction in complexity of the channel estimation process.

With reference to figure 1 a flow chart for implementation of the invention in a CDMA system such as UMTS is shown as comprising five steps illustrated as 1/1 to 1/5 of figure 1. The BS includes in its transmissions a word in the control channel which provides an indication of the transmitted power level at the instant of the transmission. As currently proposed for the UMTS FDD mode specifications the indication of transmitted power at the instant of transmission forms part of the calculation of the initial up-link transmission power for the random access channel. step 1/1 The MS measures the received power level and by means of a power loss formula estimates the distance from BS to MS. A power loss formula suitable for use in this step may be found in e. g. UMTS 30.03 published by the European Telecommunications Standards Institute. step 1/2 The delay spread window is set initially using a window size and offset in accordance with the estimate for MS to BS distance obtained in the previous step 1/1. step 1/3 The actual delay spread (delay profile) at the receiver of the MS is measured using the channel estimator. step 1/4 A further estimate of the signal path length from BS to MS is computed using path loss calculations as in step 1/2 above. step 1/5 The delay spread window size is set according to the estimated MS to BS distance modified by reference to the current measured delay spread.

The sequence of steps 1/3,1/4 and 1/5 is then continued and the window size adapted continuously until communication in the channel has finished.

As the environment in which the MS is used will not normally be known, a general model may be chosen to provide from the path loss calculation a worst case estimate of the distance from the BS. The path loss model for the vehicular test environment provides the largest BS to MS distance estimate for a particular path loss estimate. This worst case estimate ensures that there is some confidence that the setting of the corresponding delay spread window is large enough to encompass all likely significant multipath signal energy.

By way of a further example to demonstrate the methodology, a decision to switch between two sizes of search window is based on an estimate of the distance between the MS and BS. Referring once more to the steps 1/1 through 1/5 and two sizes of search window, Step 1/1 estimate the distance from the MS to the BS. Typically (as proposed in UMTS 30.03) an antenna height of 15 metres may be assumed with a carrier frequency of 2 GHz (appropriate for the UMTS down-link carrier frequencies). This yields a signal path distance formula, for the vehicular test environment, of: shortest signal path in Kilometres = lo [ (path loss-128.1)/37.6] where"path loss"is the difference in decibels established by calculating the difference between the signal power at the instant of transmission, as determined from the received broadcast channel, and the received signal power measured at the receiver.

With reference to figure 3 a graph giving estimates of path loss versus shortest signal path is shown. To allow for the effect of shadow fading a standard deviation of lOdB may be assumed. It may also reasonably be assumed with E high degree of probability that the effect of shadowing introduces a variation of no more than 3 times the standard deviation (+/-30 dB) to the path loss value predicted by the above formula.

From figure 3 it can be seen that a measured path loss of 110 dB plus 30 dB of allowance for the effect of shadowing to give a total path loss of 140dB corresponding to an estimated distance between BS and MS of 2 kilometres.

The decision criterion is then: -If the path loss estimate is less than or equal to llOdB, the MS is less than 2 kilometres from the BS -otherwise the MS is greater than 2 kilometres from the BS step 1/2 Initialise the delay spread window size. The maximum delay spread window size is +/-16 microseconds which is comparable to the maximum delay spread for the vehicular channel impulse response model described in UMTS 30.03.

This maximum delay spread corresponds with a 9.6 kilometre difference between the shortest and longest paths. In a Rayleigh or Rician fading environment, those multipath reflectors associated with received signal paths contributing a significant percentage of the received signal energy are typically local to the MS. Therefore it may be concluded that if the MS is within 2 kilometres of the BS then it is unlikely that there exists any important signal path whose length is significantly greater than a factor N times the estimated MS to BS distance. A preferred value for N may be determined empirically but we shall consider N=2 as an appropriate value for this example. This value N=2 gives a maximum path length difference of 4 kilometres, and a corresponding delay spread window of +/-7 microseconds.

The decision criterion is then: -If the MS to BS distance estimate is less than 2 kilometres then set the delay profile window size to be +/-7 microseconds -otherwise set the delay profile window size of +/-16 microseconds.

Step 1/3 Perform the delay profile measurement with the window size determined from the previous step, step 1/2. Calculate the received signal delay spread (the time difference between the first and the last received multipath components). To minimise the variance of this estimate between subsequent cycles of the method, averaging of the received signal delay spread may be performed over several iterations.

Step 1/4 Re-calculate the MS to BS distance in the same manner as in step 1/2.

Step 1/5 Set the delay spread window size. The setting of the delay spread window is in accordance with the calculated received signal delay spread from step 1/3 multiplied by a factor M. The factor M is a confidence factor to produce a size of delay spread window sufficient to accommodate the greater part of the group of significant signals arriving at the receiver. A suitable value for the confidence factor M may be obtained by experiment in similar manner to the parameter N supra. We shall consider here a value assigned to M of 1.5.

A calculated received delay spread of +/-6 microseconds would therefore require a window of +/-9 microseconds. The minimum window size, assuming a MS to BS distance less than 2 kilometres is +/-7 microseconds.

The delay spread window under these conditions will be set to +/-9 microseconds.

A calculated received delay spread of +/-4 microseconds would require a window of +/-6 microseconds. As the minimum window size required by the distance estimate is +/-7 microseconds, however, the window size would remain at +/-7 microseconds. The MS to BS distance estimate then becomes a minimum setting for the delay spread window. This minimum setting is necessary to accommodate a rapid change in the actual received signal delay spread after one or more received signal delay spread calculations are less than might typically be expected for the MS environment.

It will be apparent that the setting of the window size is not limited to the 2 switched window sizes described and the number and sizes of window settings may be varied. In general, decisions to switch from one fixed window size to another fixed window size are implemented at threshold estimates of the shortest signal path length from transmitter to receiver computed within the receiver In a TDMA system such as GSM, the timing advance established at the BS and transmitted routinely to the MS can be used directly to provide an estimate of shortest signal path. With reference to figure 2, there is shown a flow chart for implementation of the invention in a TDMA system such as GSM comprising five steps as illustrated in 2/1 to 2/5 of figure 2. step 2/1 Estimate the MS to BS distance from the timing advance signal. step 2/2 Initialise the channel estimate window in accordance with step 2/1. step 2/3 Perform channel estimation e. g. by correlating the expected burst mid-amble sequence with the received burst mid-amble samples. step 2/4 Recalculate MS to BS distance as in step 2/1. step 2/5 Modify the channel estimation window size by reference to the current measured channel estimate spread and the calculated MS to BS distance.

Implementation of an adaptive delay spread window as described above may cause occasional exclusion of a significant signal having a path length longer than the worst case situation adopted. Exclusion of the longer path length signals will cause some degradation in performance of the receiver. The effect of the exclusion of such signals can be taken into account by including an error margin when setting the size of the delay spread window. The size of the error margin may be traded off against the increase in complexity arising from the increased size of the delay spread window when including the error margin.

Claims

Claims 1. A delay spread estimation device for a receiver in a digital radio system in which the size of a delay spread window and start time of search is set in accordance with an estimate of the shortest signal path length from transmitter to receiver computed within the receiver and modified in accordance with current measures of delay spread and re-estimates of shortest signal path length.
2. A delay spread estimation device as in claim 1 in which estimates of shortest signal path lengths are computed in the receiver by a path loss calculation.
3. A delay spread estimation device as in claim 2 in which the path loss calculation is computed from the equation: distance = Io [(loss-128. 1)/37. 6]
4. A delay spread estimation device as in any preceding claim in which a current measure of delay spread is an average value derived from a number of individual measurements.
5. A delay spread estimation device as in claim 1 in which estimates of shortest signal path lengths are computed by measurements in the receiver of the timing advance inserted at the transmitting station.
6. A delay spread estimation device as in any preceding claim in which decisions to switch from one fixed window size to another fixed window size are implemented at threshold estimates of the shortest signal path length from transmitter to receiver computed within the receiver.