CHANNEL ALLOCATION SYSTEM
This invention relates to channel allocation systems, suitable for use in cellular mobile radio telephone systems. In a cellular mobile telephone system individual mobile telephones communicate with the fixed part of the network, specifically the radio base stations, over radio links. Each radio link has a given frequency, which is allocated to the relevant radio base station. The same radio frequency is also used by other radio base stations; the frequency re-use pattern, that is the allocation of frequencies to base stations, being such that co-channel interference is minimised. A radio base station may have more than one frequency allocated to it. However, the available number of frequency channels is limited. In areas of low traffic demand a base station may have only one or two frequencies allocated to it whereas a base station in an area of higher demand would require a larger number of frequencies to be allocated. Higher demand may also be accommodated by providing base stations at more closely spaced locations, having lower power outputs to minimise the effective range of each base station so that co-channel interference is still avoided.
A voice signal requires less bandwidth than that provided by a frequency channel, and therefore most cellular radio systems use time division multiplex systems to allow several mobile units to communicate with a base station on the same radio frequency. For example, in the well established GSM (Global System for Mobile Radio) digital system eight voice channels can be carried on each frequency. By defining a sequence of 'frames', each frame being divided into eight time slots, each mobile unit can be allocated to a logical channel using one such slot in each frame. Digitisation of the speech signal allows it to be compressed so that it can be accommodated in the allotted timeslot. In the GSM system each frame is 4.6 milliseconds in duration, and each of the eight timeslots in a frame comprises 148 digital bits. Radio transmissions are prone to a number of degradations caused by the propagation characteristics of radio waves. In particular, co-channel interference (that is interference between two signals using the same radio frequency) cannot be entirely avoided, however good the frequency reuse pattern is. For example, a mobile unit at an elevated position (for example in a tall building or on a hill) may
be within range of more than one base station operating at the same frequency, because the base stations have to have sufficient power to allow reception by mobile units within their notional range of coverage which do not have a direct (line-of-sight) path to them. As shown in Figure 1 , a mobile unit MS1 operating to a base station BS1 but also in range of a base station BS2 operating on the same frequency will detect signals transmitted from base station BS2 intended for other mobile units e.g MS2, as well as those transmitted from the base station BS1 . Conversely the base station BS2 will detect interference from the mobile station MS1 , as well as the signal intended for it, which is being transmitted from the mobile station MS2.
Other problems can arise through "fading", which is a phenomenon caused by multipath interference (that is interference between radio signals which travel over two or more different paths between the transmitter and receiver. In many cases one of these paths may be the direct " ne-of-sight" path from the transmitter Other paths are generated by reflection of the radio waves off objects such as buildings, mountains, vehicles or vegetation. The differences in path length cause constructive and destructive interference patterns. Since the locations of the zones of constructive and destructive interference are dependent on the radio frequency, the locations of the fades (zones of destructive interference) will differ with frequency For this reason the existing GSM system uses a system known as "frequency hopping" in which a mobile unit does not communicate with its base station on a single frequency throughout a call, but switches from one frequency to another according to an established pattern, which may either be cyclic or pseudo-random. (A "pseudo-random" pattern is one which is cyclic over a long period, typically several hours.) This "hopping" ensures that any frequency- dependant degradation of the signal is shared out equally among all the users, instead of being concentrated on those users which happen to have been allocated to the frequency on which the problem arises.
Frequency hopping cannot be used in all situations. In a few cases, a base station may have only one frequency allocated to it. A more general problem is that one of the frequencies allocated to each base station carries a control channel (known as the BCCH) which is used for setting up calls to and from mobile units, and in particular for maintaining communication with mobile units which have not currently got a call in progress. The BCCH allocated to each base station is
permanently allocated to a timeslot (timeslot 1 ) at a given frequency. This is necessary so that a mobile unit can establish initial contact with the base station in question. In particular, the mobile unit must have information as to the frequency on which the BCCH of the local base station operates so that it can receive synchronisation instructions, and be directed as to which frequency and timeslot the mobile unit is to communicate on. When a mobile unit is in communication with one base station, the BCCH of that base station also transmits details of the BCCH allocated to each neighbouring base station so that the mobile unit can listen out for these neighbouring base stations' BCCHs and arrange a handover if appropriate.
Because the BCCH is permanently allocated to one frequency, any frequency hopping pattern must avoid moving the BCCH to another frequency or moving a mobile unit to the BCCH. If the frequency hopping pattern excludes the frequency on which the BCCH is carried altogether, then all other logical channels (i.e. the other timeslots) on the frequency will not hop either. Alternatively, the frequency hopping pattern used for timeslot 1 must be different from that used for the other timeslots, since there is one fewer frequency available, again in order to avoid the BCCH being involved in the frequency hopping pattern. In particular, in a base station having only two frequencies, any mobile unit allocated to timeslot 1 will not be able to 'hop' between frequencies at all, because it could only exchange with the frequency which is permanently allocated to the BCCH.
According to the present invention there is provided a mobile telephone system comprising a plurality of base stations and a plurality of mobile units capable of radio communication with the base stations, each base station being allocated at least one radio frequency for radio communication with one or more of the mobile units, using a time division multiplex system, the time division multiplex system having a repeated frame structure comprising a plurality of timeslots, wherein the system is arranged such that at least one of the base stations exchanges signals with mobile units in different timeslots in successive frames, and at least one timeslot and frequency being reserved for a control channel, such that no mobile unit operating on a reserved frequency is allocated to a reserved timeslot.
According to a second aspect, there is provided a method of operating a mobile telephone system comprising a plurality of base stations and the plurality of mobile units capable of communication with the base stations, in which a base station is allocated one or more radio frequencies for communication with mobile units using a time division multiplex system, the time division multiplex system having a repeated frame structure comprising a plurality of timeslots, wherein signals are exchanged between a mobile unit and a base station in different timeslots in successive frames, and at least one timeslot and frequency are reserved for a control channel such that no mobile unit operating on a reserved frequency is allocated to a reserved timeslot.
According to a third aspect, there is provided a mobile telephone for use in a mobile telephone system, having means for exchanging radio signals with a base station using a division multiplex system, the time division multiplex system have a repeated frame structure comprising a plurality of timeslots, and having means for exchanging signals with a respective base station in different timeslots in successive frames, having means for communicating with a base station on a control channel having a predetermined reserved frequency and timeslot and being arranged such that the allocated channel for exchange of radio signals does not operate on a reserved frequency and reserved timeslot simultaneously. According to a fourth aspect, there is provided a base station for a mobile telephone system, comprising means for exchanging radio signals with one or more mobile telephones using a time division multiplex system, the time division multiplex system having a repeated frame structure comprising a plurality of timeslots, and arranged such that the base station exchanges signals with individual mobile units in different timeslots in successive frames, wherein at least one timeslot and frequency are reserved for a control channel, and the sequence of available timeslots allocated for communication with an individual mobile unit is arranged such that a reserved frequency and timeslot combination are not both allocated to the same mobile unit simultaneously. In the examples to be given below, only one timeslot is reserved for use as a control channel, in only one of the available frequencies. It will be appreciated that more than one channel may be required to be reserved, for control or other purposes. These control channels each use a different timeslot/frequency pair, but
it is possible for two control channels on different frequencies to use simultaneous timeslots, and for two control channels to use the same frequency, provided they use different timeslots.
The allocation of each timeslot to a mobile unit preferably follows a cyclic pattern, which may proceed stepwise through the available timeslots in each frame or may be a pseudo-random pattern. The pattern may be combined with frequency hopping to allow hopping throughout the time/frequency domain.
By varying the timeslot used within a frame the effect of co-channel interference can be reduced, because interference between two logical channels will only occur when both channels occupy the same timeslot (as seen by the receiver, allowing for any time-lag). Co-channel interference may still occur in each timeslot if they are all in use in the interfering channel, but the interference will be from different logical channels in successive timeslots. The interfering signal will therefore be less coherent than if it came from a single logical channel, and therefore more like random noise, making it easier to extract the real signal.
The invention will now be described in detail with reference to the following drawings in which;
Figure 1 is a schematic diagram showing parts of a typical cellular radio system, illustrating co-channel interference; Figure 2 is a schematic diagram illustrating multipath interference;
Figure 3 is a schematic diagram illustrating a conventional frequency hopping system;
Figure 4 shows a simple variable timeslot pattern in conjunction with variable frequency Figures 5a to 5d show a further hopping pattern according to the invention; and
Figures 6a and 6b show another hopping pattern, providing two control channels;
Figure 7 (on the same sheet as Figure 3) shows another frequency/timeslot hopping pattern, also according to the invention.
In Figure 1 there are shown four base stations, BS1 , BS2, BS3 and BS4 and two mobile stations MS1 and MS2. Base stations are typically controlled remotely by a control system arranged to control a plurality of such base sites. The mobile station MS1 is in radio communication with the base station BS1 and
the mobile station MS2 is in communication with the base station BS2. Base stations BS1 and BS2 share a frequency f1 and both mobile stations MS1 and MS2 have been allocated to that frequency. In general the geographical locations and power outputs of the base stations BS1 and BS2 will be selected such that co- channel interference is minimised. In particular, the system will generally be designed so that base stations using the same frequency do not have coverage zones which are adjacent. For example a mobile station MS1 is, in normal use, unlikely to be able to receive a strong signal strength from both BS1 and BS2, and similarly only one of the base stations BS1 and BS2 will be able to receive a strong signal from the mobile station MS1 . Other base stations using different frequencies, e.g. base stations BS3, BS4 on frequencies f2, f3 will in general cover regions where signal strengths from base stations BS1 , BS2 would be similar. However, situations can arise when a mobile station MS1 is in radio range of two base stations BS1 , BS2 operating at the same frequency f-| . In such a situation transmissions from the mobile station MS1 will be detected at the base station BS2 and will interfere with any transmissions from the mobile station MS2, which is transmitting on the same frequency and timeslot. (Note that because of the different path lengths mobile station MS1 will not be synchronised with the base station BS2, but it will nevertheless interfere with one timeslot of the signal received at the base station BS2. Similarly, transmissions from the base station BS2 to the mobile station MS2 will be detected by the mobile station MS1 . Such situations can arise in particular where the mobile station is in direct line of sight with a base station at some distance away from it, for example on a high building or a hill, when the base station is configured for communicating with mobile stations at much shorter range but not in direct line of sight (for example in a built- up area).
Figure 2 illustrates schematically the phenomenon of multipath interference, also known as fading. The base station BS1 and the mobile unit MS1 are in radio communication with each other. It will be seen that transmissions from base station 1 can travel in a direct line of sight over a path P1 . The radio signal over this path may be attenuated by ground level objects such as foliage or topography, such that the signal received at mobile station 1 is attenuated. A second radio path P2 may exist to the mobile station 1 by way of reflections off buildings etc. This path may also be attenuated. It is also possible for both paths
P1 and P2 to be indirect. Because of the different path lengths of the path P1 and P2 the radio signals transmitted from the base station BS1 over the two paths do not arrive in phase. Constructive interference (where the path lengths differ by a whole number of wavelengths) results in the signal being reinforced, however destructive interference (where the path lengths differ by an odd number of half wavelengths) results in partial or complete loss of the radio signal. As the mobile unit MS1 moves, it will pass through areas of constructive and destructive interference. If it is stationary if may by chance be in an area of destructive interference (known as a deep fade) and the signal strength will be very much reduced. An attempt to rectify this by increasing the gain renders the mobile unit prone to co-channel interference from other base stations in the area working on the same frequency.
Because destructive interference is caused when the difference in path lengths between P1 and P2 is a whole number of half wavelengths, at any given location a change in radio frequency (and therefore wavelength) will in general result in the mobile unit no longer being in a fade. For this reason, in the mobile radio system known as "GSM", "frequency hopping" is used, in which each mobile unit switches from one frequency to another in a recognised pattern. Thus any fade in which the mobile station 1 happens to find itself will only last for one time frame, resulting in a reduced impairment in the overall signal. It is of course the case that a mobile station which finds itself in an area of constructive interference in one frame may, on the next frame, find itself in an area of destructive interference, but again this will only last for one frame before the signal is recovered. By using a prearranged frequency hopping pattern any fading will be only a very short term phenomenon. Frequency hopping has been found to overcome the problem of fading better than a reactive system in which frequencies are changed only in reaction to a fade. In particular, because the mobile station may be moving, by the time a switch to another frequency has taken place in a reactive system then the need for the change may well have passed, and indeed the change may be to a frequency for which the mobile station is entering a zone of fading.
Figure 3 shows a typical frequency hopping pattern for a base station having two frequencies available, which is the most common configuration in typical "GSM" networks. Some base stations have more than two frequencies
available, but the problem to be explained is at its greatest in base stations where only two frequencies are available. Considering a mobile station A allocated to timeslot t5, the mobile station is allocated in the first frame to a frequency f In the following frame it is switched to the other frequency f2, and it then returns in the third time frame to the frequency f-i . Another mobile unit may also operate in timeslot t5, using the frequency unoccupied by the mobile station A at any given time. Similarly other pairs of mobile units may use the other timeslots t2 to t8.
However, timeslot t-i is allocated, for frequency f-i to a base site control channel
(BCCH) which provides control information to mobile units which are not currently engaged in a full radio link with the base station. In order for mobile units to first establish contact with the base station, they need to identify the base site control channel, and for this reason the base site control channel must always remain on the same frequency. However, this means that if a mobile unit B uses this time slot t-i in the second frequency channel f2, it is unable to frequency-hop because this would require it to use the base site control channel which is already preallocated in timeslot t-| .
The mobile unit B, being unable to frequency hop, is therefore more vulnerable than the other mobile stations to signal degradation from fading, co- channel interference, and other phenomena. In the conventional GSM system the timeslot t-i (for frequencies other than that allocated to the BCCH) is allocated to mobile stations according to interference measurements from other active mobile stations, without knowledge of BCCH logical channel location or regard to frequency hopping freedom, and the frequency hopping pattern used in other timeslots is suspended or modified for this timeslot to avoid the mobile unit operating on the BCCH. This reduces the average channel quality, as logical channels allocated to timeslot t-i have fewer frequencies to hop between.
The frequency hopping pattern as shown in Figure 3 to be used by mobile station A is indicated to the mobile station when it establishes contact with the base station. The base site control channel BCCH includes, as part of the control data transmitted to the mobile stations under the control of that base station, information regarding the channel to be allocated, including the timeslot and the frequency or frequency hopping pattern.
Figure 4 shows a channel allocation scheme according to the invention. Eight successive frames of the transmissions are shown, again for two frequencies
f-i , f2. Each frame has eight timeslots ^ to t8. Again the broadcast control channel (BCCH) is permanently allocated to timeslot tt at frequency f-] in each successive frame. Both mobile stations A and B alternate between the two frequencies f-i and f2. However, unlike in the prior art arrangement they are not permanently allocated to the same timeslot. In the arrangement shown in Figure 4, in each frame the mobile station is allocated to the timeslot following the timeslot allocated to it in the previous frame. This pattern is cyclic, so that when mobile station A reaches timeslot t8 in frame 4, in the next frame (frame 5) it is allocated to the first timeslot t-, . The system is arranged such that the timeslot/frequency allocated to the BCCH is omitted in the cycle, so that base station A goes from frequency f2 in timeslot t8, to frequency f2 in timeslot t2, thereby skipping timeslot tøfrequency f-, . It will be seen, however, that when base station B reaches timeslot t8 in frame 8, using frequency f-i , it is next allocated to timeslot t-i and frequency f2 in frame 9. It will be noted that in this allocation pattern some mobile units repeat the timeslot pattern after seven frames (e.g. mobile unit A) and others after eight frames (e.g. mobile unit B).
Figure 5, (made up of four parts Figures 5a to 5d) shows a complete cycle of the channel hopping pattern of Figure 4. The left hand side shows the full channel allocation, the right-hand side is identical, but only the positions of three channels, (1 , 6, and 1 1 ) are shown, to allow their movement to be more clearly understood. The odd-numbered channels (e.g. 1 , 1 1 ) repeat every eight frames (see channel B in Figure 4), but the even-numbered channels (e.g. 6) repeat every seven frames (see channel A in Figure 4) because they skip the timeslot/frequency allocated to the BCCH. The full pattern therefore repeats every 56 frames, this being the lowest common multiple of 7 and 8 (0.2576 seconds): the first frame in Figure 5a is repeated in the last frame of Figure 5d.
Figure 6 (made up of two parts, Figure 6a and Figure 6b) shows a similar allocation pattern for a system with two control channels on radio frequency f1 ; allocated to timeslots t-, and t5. (It is usually convenient to provide all control channels on one frequency and to have them equally spaced in time, to allow them to be readily recognised, but other arrangements are also within the scope of the claims). As with Figure 5, the left hand side shows the full pattern, and the right hand side shows the movements of channels 1 ,6, and 1 1 only. In this case the even-numbered channels (for example channel 6) skip two timeslots, t^ and t5, so
they repeat after only six frames, so the full pattern repeats after 24 frames (the lowest common multiple of 6 and 8), or 0.1 104 seconds. The last frame shown in Figure 6b is the same as the first in Figure 6a.
Figure 7 shows another timeslot/frequency allocation pattern. Again the base station has eight timeslots t-, to t8, but this time the base station has five available frequencies f-i to f5. A mobile unit requiring a logical channel is allocated to a given frequency/timeslot allocation for the first frame of the transmission and then follows the slots in the numerical order shown, returning after slot 39 to slot
1 . It will be seen that in each successive frame the mobile unit changes from one frequency to the next, and from one timeslot to the next. After any frame in which it uses frequency f5 it returns in the next frame to frequency f-| , and after any frame in which it uses timeslot t8 it returns to timeslot t-i in the following frame. The exception is that after slot 1 5 (frequency f5, timeslot t8), frequency fT/timeslot ^ is omitted as this pairing is allocated to the BCCH, so the mobile station moves straight to timeslot t2/frequency f2 for slot 16.
There are prior art systems in which a pseudo-random sequence of frequency hopping is used, which repeats over several hours. This can give added security against eavesdropping. In such a pseudo-random sequence each frequency is used by a mobile unit an equal amount of time, on average. This principle may be extended to timeslot hopping as well, by allowing a mobile unit to move from a timeslot in one frame to any timeslot (including the same timeslot) in a subsequent frame in a predetermined sequence which only repeats over a very long period such as several hours. In such an arrangement the pseudo-random sequence and also the position in the pseudo-random sequence must be identified to the mobile unit by the base station, so that the mobile unit can follow the pattern The pseudo random sequence of the timeslots may have a different repeat period to the pseudo random sequence of the frequencies, so that the whole pattern repeats very rarely indeed. The number of frames in a full repeat is the lowest common multiple of the number of frames in each of the frequency and timeslot repeat patterns. For example, if the frequencies and timeslots each repeat over a period of about an hour, such that the number of frames in each pattern is approximately 800,000 frames, but selected to have no common factors, (e.g. 799,995 and 800,008 frames respectively), the complete pattern will repeat in
approximately 640,000,000,000 frames, or 800,000 hours, which is more than ninety years.
It should be noted that even in the simple timeslot changing systems shown in Figures 4 to 7 the arrangement is not equivalent to simply changing the length of the frame. The mobile units' operation varies with respect to the fixed reference points of the frame such as the BCCH, as well as the guard periods and other control bits (which are not shown in the diagram, for simplicity).
By varying the timeslot on which each mobile unit is operating, as well as the frequency, each logical channel has all frequencies allocated to the base station available to it. Thus no mobile stations are permanently allocated to timeslot t-i , which would require that they have one of the available frequencies permanently barred to them for the duration of the call. Furthermore, should a mobile stations' transmissions be detectable at a base station other than the one to which it is supposed to be operating, the timeslots with which it will interfere will be different in each frame, and will therefore, in general, interfere with different mobile stations in each frame, thereby reducing the effects of the interference on any given logical channel. Different timeslot hopping patterns may be used at each base station, in order to avoid accidental synchronisation of logical channels at different base stations. Conversely, if a mobile unit can detect transmissions from more than one base station, the interfering signal which is synchronised with an individual timeslot of the logical channel to which the mobile station is allocated will relate to a different logical channel at the interfering station for other timeslots. Thus although an interfering mobile unit will be detectable, each successive timeslot of the interfering signal will relate to a different logical channel so the interfering signal will be mere noise and not a coherent speech (or data) signal and will thus be less intrusive.