WO2008115105A1

WO2008115105A1 - A method and a device for reduced interference in a cellular access system.

Info

Publication number: WO2008115105A1
Application number: PCT/SE2007/050165
Authority: WO
Inventors: David Astely; Tobias Tynderfeldt; Erik Westerberg
Original assignee: Telefonaktiebolaget Lm Ericsson (Publ)
Priority date: 2007-03-16
Filing date: 2007-03-16
Publication date: 2008-09-25

Abstract

A method (700) for use in a cellular system (200) comprising a plurality of base stations (112, 122, 132, 142) for the control of the traffic to and from a cell (110, 120, 130, 140), in which system (200) the communication between users in a cell and their base station is divided into an up link period (UL), and a down link period (DL), with guard periods between uplink and downlink periods. Each base station utilizes time information from one of a number of synchronization sources in order to time the down link, up link and guard periods properly. The method comprises letting each of the base stations use information on the inaccuracy of the synchronization source of the base station in order to adapt its downlink periods so that they do not interfere with up link periods in one or more of the other base stations in said plurality.

Description

TITLE

A method and a device for reduced interference in a cellular access system.

TECHNICAL FIELD The present invention discloses a method for use in a cellular access system in which there is a number of base stations, each of which controls the traffic to and from at least one cell in the system.

In the system in which the invention may be applied, communication between users in a cell and the base station of the cell is divided into an up link period during which users may transmit to their base station, and a down link period during which a base station may transmit to users in the cell of the base station. There are also guard periods between the uplink and downlink periods, and the base stations utilize time information from one of a number of synchronization sources in order to properly schedule the down link, up link and guard periods properly.

By means of the invention, improved performance and robustness with regard to interference in the system can be achieved.

BACKGROUND

In the LTE (Long Term Evolution) of 3GPP, a new physical layer based on OFDM (Orthogonal Frequency Division Multiplex) in the downlink and SC- FDMA (Single Carrier Frequency Division Multiplex Access) is currently being standardized. This physical layer should support both FDD (Frequency Division Duplex) and TDD (Time Division Duplex) operation, and there should be a high degree of commonality between these two modes of operation.

In TDD systems, the communication between users in a cell and the base station of the cell is divided into an up link period during which users may transmit to the base station, and a down link period during which base stations may transmit to users. Thus, in a TDD system, as compared to FDD systems, interference between up link and down link transmissions needs to be considered since both the up link and the down link use the same frequency resources.

A terminal transmitting uplink to a base station may cause interference in another terminal which is receiving in down link in another cell. Similarly, a base station transmitting down link to a terminal may cause interference in another base station which is in the uplink mode, i.e. is receiving data.

One solution to the problem of interference between up link and down link transmission is to synchronize and coordinate the base stations in question, so that all uplink and downlink periods occur simultaneously in all cells.

In addition to the synchronization and coordination of up link and down link periods, so called guard periods may be inserted at the transitions from down link to up link and at the transitions from up link to down link.

Typically, the guard time between down link and up link periods is chosen to match the maximum roundtrip propagation delay in a cell, in addition to the time it takes for a user terminal to switch from reception to transmission. This is due to the fact that the propagation delay causes a delay before a terminal can receive downlink data. In addition, the transmission timing of the terminals is controlled so that the terminals will start their up link transmissions at a point in time which compensates for the propagation delay, so that the transmissions are within the uplink "window" at the base station.

The guard period between the uplink and downlink periods, on the other hand, is typically chosen to match the time it takes for a base station to switch from reception to transmission and for a (nearby) user terminal to switch from transmit mode to receive mode. As can be realized, proper synchronization of the base stations is essential in a TDD system if interference between uplink and downlink is to be avoided in the way described, and for this reason, a time reference or synchronization signal must be provided. There are several different alternatives for this, such as, for example, a satellite based system such as the GPS system, which is a solution which may be suitable for an outdoor base station, but which may be less well suited for a base station indoors, which may have poor GPS coverage.

Other solutions comprise providing synchronization signals to the base stations via the transport network, possibly by means of IP signalling, which will typically provide poorer accuracy than GPS as well as "over-the-air" synchronization methods, which are part of the air interface of the base stations, and which typically needs to be supported by the standard and may require a different network deployment, since each base station may need to be able to listen to its neighbouring base stations, as well as the fact that user terminals may need to be able to hear several base stations with sufficient quality simultaneously.

An additional problem is that the base stations may temporarily lose synchronization, for example due to hardware failure or due to low traffic, which can be the case if. for example, user terminals are involved in the over- the-air synchronization. There is also a possibility that different base stations use different time references.

In order to address the problem of inter base station interference caused by poor synchronisation by means of adapting the guard periods to this, the guard periods would need to be large enough to account for the synchronization errors, which might have a heavy impact on system performance, since no transmissions may be made during the guard periods. Additionally, if synchronisation is lost, the impact of inter base station interference may start to degrade performance, since very high levels of interference may occur during parts of the uplink period. Similarly, there may also be interference between terminals caused by this.

SUMMARY

Thus, as explained above, there is a need for a solution by means of which the problem of base station synchronisation inaccuracy may be solved in a manner better than solutions known at present.

This need is addressed by the present invention in that it discloses a method for use in a cellular access system which comprises a first plurality of base stations, each of which controls the traffic to and from at least one cell in the system.

In the system in which the invention may be applied, communication between users in a cell and the base station of the cell is divided into up link periods during which users may transmit to their base station and down link periods during which a base station may transmit to users in the cell of the base station.

The system in which the invention may be applied also comprises guard periods between uplink and downlink periods, and each of the base stations utilize time information from one of a number of synchronization sources in order to properly time the down link, up link and guard periods.

The method of the invention comprises letting each of the base stations in the plurality of base stations use information on the inaccuracy of the synchronization source of the base station in order to adapt its downlink periods so that they do not interfere with up link periods in one or more of the other base stations in the plurality. Suitably, according to the invention, the down link periods are adapted by shortening them and shifting them in time by delaying their start.

Preferably, the down link periods are shortened by silencing one or more symbols at the end of a down link period, although the same may be done at the beginning of a down link period.

These and other embodiments of the invention will be described in more detail below.

The invention also discloses a base station for carrying out the method of the invention.

BRIEF DESCRIPTION OF THE DRAWINGS The invention will be described in more detail in the following, with reference to the appended drawings, in which

Fig 1 shows a problem intended to be addressed by the present invention, and Fig 2 shows a prior art solution, and

Fig 3 and 4 show another problem to be solved by the invention, Figs 5-8 show various embodiments of the invention, and Fig 9 shows a block diagram of a base station of the invention, and Fig 10 shows a flow chart of a method of the invention.

DETAILED DESCRIPTION

Fig 1 schematically shows a system 100 in which the invention may be applied. The system 100 is a cellular wireless access system, and as such comprises a number of cells, two of which are shown in fig 1 with the reference numbers 110 and 120, respectively. The cell 110 comprises at least one radio base station, an RBS, shown as 112 in fig 1. The RBS 112 serves, inter alia, to control the traffic to and from users in the cell 110. The cell 110 can accommodate at least one user with a user equipment, UE, shown in fig 1 with the reference number 114. Similarly, the cell 120 is shown as comprising an RBS 122 and a UE 124.

The system 100 is shown as a cellular telephony system, and the invention will be described with reference to such a system, but it should be pointed out that this is by way of example only, the invention may be applied to a number of different kinds of cellular wireless access systems.

Also, the terminology used when describing the invention with reference to the system 100 is merely intended to facilitate the reader's understanding of the invention and is not intended to restrict the scope of protection sought for the present invention. For example, the term "base station" or "radio base station", RBS, should be interpreted as meaning a node in the system with the function of an RBS. In some systems, for example, a function essentially corresponding to that of the RBS is carried out by a node in the system called Node B. Such systems are naturally also encompassed by the invention.

Similarly, it should be realized that the term user equipment or UE is merely an example intended to facilitate the reader's understanding of the invention. For example, in some systems, the terms UT, User Terminal, or MS, Mobile Station, are used. Naturally, such systems are also encompassed by the scope of the present invention.

In addition, although the UEs 114 and 124 are shown as cellular telephones in fig 1 , it should be realized that this is merely to facilitate the understanding of the invention, the UEs may be many other kinds of devices, portable or stationary, such as, for example, computers. As explained previously in this text, the UEs 114 and 124 can send traffic to the RBSs 112. 122, as well as being able to receive traffic from the RBSs. This is shown by means of arrows between each of the RBSs and its respective UE. A problem in systems of the kind 100 shown in fig 1 is that transmissions from one RBS, which are thus intended for the UEs in the cell of that RBS, may be received by other RBS in the system 100.

Thus, inter-RBS interference is indicated in fig 1 by means of an arrow between the RBSs 122 and the RBS 112. Similar problems may occur between the UEs, as indicated by an arrow between them. As mentioned initially, one way of avoiding such interference in a TDD system, Time Division Duplex, is to synchronize the system and to coordinate the uplink and downlink periods so that they occur simultaneously in all cells in the system.

So called guard periods may be inserted at the transition from UL to DL₁ and at the transition from DL to UL. The guard periods are intended to take into account a variety of effects such as, for example, switching times from send to transmit and vice versa, propagation delays in the system etc.

The principle of guard periods at the transitions from down link to up link, and also at the transition from up link to down link is shown in fig 2 for the RBSs 112 and 122 from fig 1. as well as for an additional RBS 132 from another cell in the system 100. The guard periods between DL and UL are denoted as T_0U, and the guard periods between UL and DL are denoted TUD-

If the guard periods T_Du and/or TDU are too short, problems with interference will be encountered, while, on the other hand, if the guard periods are too long, this will result in a degraded system performance, since less time is available for data transmissions due to the use of guard periods. In order to avoid interference between uplink and downlink transmissions by means of coordinating the transmission of the cells in a system, it can be realized that synchronization between the RBSs in the system is required. Fig 3 schematically shows what might happen in the system of figs 1 and 2 if the synchronization inaccuracy of the RBSs in the system becomes too high: as indicated in fig 3, the RBS 122 has lost its synchronization, which has caused it to initiate the entire sequence of DL and UL periods later in time than intended, and RBS 132 has a synchronization problem which causes it to initiate the sequence earlier than intended.

One effect of this synchronization inaccuracy is indicated in fig 3: the DL periods of RBS 122 overlap the UL periods in RBSs 112 and 132, and the DL transmission of RBS 122 may thus interfere with the UL reception in RBSs 112 and 132.

The problems of synchronization error between the RBSs in a system are further illustrated in fig 4, where an additional RBS 142 has been inserted. The RBSs 112 and 122 are "in synch", while the RBSs 123 and 142 are "out of synch". RBS 132 is out of synch by an amount 2Δ, while RBS 142 is out of synch by the same amount -Δ. Thus, RBS 132 has a positive synch error, while RBS 142 has a negative synch error.

For a positive synchronization error, such as with RBS 132, this RBS is interfered at the end of its UL period, and for a negative synchronization error, such as the one of RBS 142, this RBS is interfered at the beginning of its UL period.

Thus, a purpose of the present invention is to alleviate the problems caused by RBS synchronization inaccuracies. A basic principle according to the invention is to let the base stations involved adapt their respective DL periods so that the base station^'s DL transmissions do not interfere with other base stations, especially with the UL transmissions in the cells of the other base stations. According to the invention, one way of achieving this is to let a base station compensate the guard period at the transition from DL to UL for synchronization inaccuracies in the sync source of the base station.

This compensation is obtained by letting a base station reduce its DL transmission periods and to shift them in time based on the inaccuracy of the base station's synch source, or on an estimate of the synch source's inaccuracy, so that the DL transmissions of one base station do not interfere with UL transmissions in other base stations. In effect, an increased guard period at said transition is obtained. An alternative is to silence entire DL sub- frames.

In this way, base stations which have different synchronization accuracy, usually caused by the use of different synchronization sources can co-exist in the same area with no base station to base station interference. As will be seen in the following, at least a temporary loss of synchronization in one or more base station can be handled by means of the invention.

In order to carry out the compensation for synchronization inaccuracy, the inaccuracy of the synch source needs to be estimated by the base station. A base station can use one or more sources for synchronization, each of which source may have different inaccuracies. Examples of synch sources available to a base station are satellite based systems such as the GPS system, synchronization over the transport network to which the base station is connected. Alternatively, synchronization can also be carried out "over-the- air", for example by providing synchronization signals between base stations or by using measurements from user terminals that are in the coverage area of the system.

Associated with each synchronization source or method, an expected accuracy or inaccuracy can be assigned. The base stations may continuously monitor the status of each of the synchronization sources, and based on which source the base station uses determine an expected inaccuracy of the timing of the base station.

If the availability of one synch source is lost, the base station can select a second, usually less accurate, source, and update the figure for the synchronization accuracy accordingly.

Additionally, for example if all external synchronization sources are lost or temporarily unavailable, a base station can use the expected drift/stability of an internal clock source in the base station, and from this determine an expected synchronization error as a function of the period of time that the internal clock is used as a synch source.

Thus, a base station in a system of the invention will be able to determine the expected inaccuracy of the synch source, and accordingly it will also be able to determine the inaccuracy of the radio frame synchronization.

Using the determined inaccuracy of the synch source, the base station can then act on the inaccuracy, if it exceeds a certain predetermined limit.

According to the invention, each base station may autonomously increase the guard period T_Du at the transition from DL to UL, which is done by introducing a number of "idle" or silent DL symbols in the last sub frame of each DL period. As an alternative, entire sub frames or slots can be set to be idle.

As an example of this principle, suppose that the expected synchronization inaccuracy of the synch source of a particular base station is determined to be in an interval [-Δ, Δ] relative to the common reference synch used in the system. In order to compensate for this, the guard period T_DU in the base station is increased by 2Δ. in order to handle the expected synchronization error, an increase which is obtained in the manner described, for example by silencing symbols at the beginning or end of the DL periods of the base station. In addition to this increase in guard period, the entire send/transmit sequence, i.e. the whole "radio frame" of the base station needs to be delayed, suitably by half of the guard period increase, i.e., Δ in order to avoid any overlap between uplink reception and downlink transmissions in the base stations.

This procedure is illustrated in Fig 5. for the case when RBS 122 happens to have a synchronization error in the interval of ±Δ, and RBS 112 is "in synch", or at least has a synchronisation error which is below a predefined tolerance limit. The DL and UL periods of the RBSs are illustrated with arrows which point down and up, respectively.

As can be seen in fig 5, the DL periods of RBS 122 have been shortened by 2Δ, and in addition, the entire radio frame 500, i.e. the send/transmit sequence of RBS 122 has been shifted Δ in time, in this case by being delayed by Δ.

In fig 6, a more general case is illustrated in which the invention is applied. An RBS can only calculate or estimate an interval [± Δ] for the inaccuracy of its synch source, as opposed to being able to calculate an exact figure for the inaccuracy, or the "sign" of the inaccuracy, i.e. if the inaccuracy Δ is positive or negative.

Thus, a measure taken by the RBS to compensate for the synch inaccuracy must take into account an inaccuracy in an interval [± Δ], Δ being the calculated value of the synch inaccuracy. This is shown in fig 6, where synch inaccuracy of RBS 112 is below a predetermined threshold, i.e. RBS 112 is "in synch", while the RBSs 122, 132, 142, have determined that their respective synch inaccuracies are above the acceptable value. For the sake of simplicity, it will be assumed in the example of fig 6 that all of the inaccuracies of these RBSs are in the same interval, i.e. [± Δ]. This, however, need not be the case.

The actual case is that RBS 122 by chance is in synch, although the sync inaccuracy is above the acceptable threshold, while RBS 132 is delayed by Δ, i.e. has an actual error of +Δ, and RBS 142 is too early in time by Δ, i.e. has an actual error of -Δ. These actual errors are, however, as explained, unknown to the RBSs, but are explained here so that the reader will more easily understand the effects of the invention.

In the example shown in fig 6. each of the RBSs 122. 132, 142, has increased the guard period T_Du between its DL and UL periods by 2Δ, i.e. by the entire span of the calculated inaccuracy, while the entire radio frame 622, 632, 642, of each of the RBSs has been shifted in time by introducing a delay of Δ.

As can also be deduced from fig 6, as well as from fig 5, the method shown in those figs will serve to avoid interference between the base stations at the beginning of the UL periods, at least if said interference is caused by synch inaccuracies in the RBSs. However, there may still be some interference at the end of the UL periods of some of the RBSs in the group, as can be seen for example by an overlap of the final DL period in fig 6 of RBS112 and the UL period of RBS 122. This problem too, may be addressed by the present invention, as will be described below. However, it may be seen that this problem is experienced by those base stations which are out of sync, and not by the majority of base stations, which are expected to be in sync.

In one embodiment of the invention, the solution shown in fig 5 or 6 is combined with the following in order to reduce or eliminate the problem of interference in the UL periods. This embodiment is shown in fig 7, in which the same RBSs as in fig 6 are shown, and the embodiment can be described as follows: The RBS which suffer from interference in their respective UL periods shorten their UL periods by an appropriate length of time, which may be determined by repeated measurements or iterations to see how much the UL period needs to be shortened in order to reduce the interference to an acceptable level.

The UL periods can be shortened by, for example, introducing an appropriate number of idle symbols at the beginning or at the end of the UL period, or by silencing an entire sub frame, basically in the same way as for the DL periods. Typically, a reduction of the UL period by 2Δ in the RBSs with poor synchronization accuracy will be sufficient, which is what is shown in fig. 7.

Thus, in fig 7, as compared to fig 6, the UL periods are shortened by 2Δ, which is in addition to the time shifting and shortening of the DL periods, which has been described in connection to fig 6.

In another embodiment of the invention, in which it is also desired to reduce interference in the UL periods as well as in the DL periods, the following approach, the results of which are shown in fig 8, is instead utilized:

Based on interference measurements in the UL periods, the RBSs will move their UL "window", i.e. the receive window, and will use so called "timing advance" in order to control their UE's transmission timing to align with this. Timing advance is used by an RBS to inform a UE of the move of the receiver window, so that the UE may adjust its transmission timing accordingly.

Using this embodiment, the UL periods do not need to be shortened, as can be seen in fig 8. As can also be seen in fig 8, by means of the embodiment of that fig, the UL periods do not need to be shortened, since they are instead advanced (or retarded) in time, which is done in addition to the time shifting and shortening of the DL periods..

Thus, the DL periods are shortened by the RBSs using information on the inaccuracy of the synchronization source used by the individual RBS. As opposed to this, the UL periods can be left "as they are", or they can be shifted in time or shortened.

Turing now to the means necessary in an RBS in order to carry out the method if the invention, this will be described together with fig 9, which is a block diagram which shows some of the components in an RBS such as the RBS 112 shown previously, equipped according to the invention.

Thus, the RBS 112 of the invention will comprise means for utilizing time information from one of a number of synchronization sources in order to time its DL, UL, and guard periods TDU, TU_D, properly. These means will suitably be comprised in the receiver part 910 of the RBS, as well as in a calculating part of the RBS, such as a microprocessor 914. Thus, the receiver part 910 receives synchronization information from the synch source, and the calculating part 914 calculates the synch accuracy or inaccuracy associated with each synchronization source. In the case that has been mentioned above, where an internal clock is used as synch source for the RBS 112, the receiver naturally does not come into the synch processing.

Further, in the case that over-the-air synchronization is used, the RBS may contain means for transmitting and receiving synchronization signals or measurements from one or several terminals or base stations, and means for processing them in order to determine a timing estimate as well as an expected accuracy. Suitably, these means are also comprised in the calculating means 914 and the transmit and receive means 910, 912. The calculating part 914 can accesses a memory 916 where there is information stored on the accuracy of each of the possible synch sources for the RBS 112, and uses this information in order to calculate the synch accuracy or inaccuracy.

The inaccuracy of the synch source which is calculated by the calculator 914 is then evaluated, suitably also by the calculator 914, against a predefined threshold to see if the DL periods of the RBS 112 need to be adapted in order not to interfere with up link periods in other RBSs.

If the DL periods need to be adapted, this is suitably carried out by means of shortening them and shifting them in time by delaying their start. The shortening can also be carried out by the calculator 914 of the RBS 112, which calculates how much the DL periods need to be shortened, and silences symbols or entire sub frames in order to achieve this. In addition to shortening the DL periods, they are shifted in time by delaying their start, following which they are transmitted by a transmitter function 912 in the RBS 112.

If the synchronization source is an external one, the RBS's 112 interface towards the synch source is the receiver 910, and the external synch source may be one of the following:

• Signals from a satellite based system, such as the GPS system,

• Synchronization signals from a transport network to which the system is connected,

• Synchronization over the air interface, by means of exchanging synchronization signals and measurements between base stations and terminals in the system.

The synch signals are thus received by the receiver 910, and processed by the calculator or processor 914. The processor 914 may also carry out interference measurements or measurements related to interference such as throughput, quality or the total power received in the UL periods of the RBS 112, by means of signals received by the receiver 910. If the UL interference exceeds a certain predefined level which defines what is acceptable, the processor 914. together with the transmitter 912 may shorten the UL periods, in order to avoid interference.

Suitably, this is done by shortening the UL period at its beginning or end, by means of silencing one or more symbols during the last part of the up link period or by means of silencing entire UL sub frames.

In addition, the processor 914 together with the receiver 910 may initiate the UL link period earlier or later in time than the nominal starting point of the up link period, and together with the transmitter 912 inform users in the cell of the RBS 112 of the new starting time of the up link period.

Fig 10 shows a schematic flow chart 1000 of a method of the invention. Steps which are alternatives or options are shown with dashed lines. As shown in fig 10, in step 1010 each of the base stations in the plurality of base stations to which the invention is applied uses information on the inaccuracy of its synchronization source in order to adapt its downlink periods so that they do not interfere with up link periods in one or more of the other base stations in said plurality.

As indicated in step 1020, one way of adapting the DL periods is to shorten them and to shift them in time by delaying their start. Steps 1030 and 1040 show that the DL periods can be shortened either at their ends or at their beginnings. As indicated in step 1050, another option is to let the base stations shorten their UL periods.

The invention is not limited to the examples of embodiments shown in the drawings and described above, but may be freely varied within the scope of the appended claims.

It can for example be pointed out that within the scope of the invention, the up link period may be initiated earlier or later in time, based on measurements of a performance indicator representing the interference level, said measurements being carried out in the time interval between two consecutive down link periods.

Also, an RBS in a system of the invention may be equipped with means for informing users in its cell of changes in the duration and/or starting point of the downlink periods.

Finally, the synchronization source used by an RBS of the invention may be chosen from among a number of synch sources available to the RBS, in which case the choice is suitably carried out by a function for this in the RBS. In such a case, the choice is made, for example, based on known accuracies for the various sources, with the most reliable source usually being chosen by said function.

It should also be pointed out that the expression "a number of synchronization sources" is intended to cover the case where only one source is available to an RBS, due, to for example, a choice made when installing the RBS.

Claims

1. A method (1000) for use in a cellular access system (200), said system comprising a first plurality of base stations (112, 122, 132, 142), each of which controls the traffic to and from at least one cell (110, 120, 130, 140) in the system, in which system (200) the communication between users in a cell and the base station of the cell is divided into an up link period (UL) during which users may transmit to their base station, and a down link period (DL) during which a base station may transmit to users in the cell of the base station, the system also comprising guard periods between uplink and downlink periods, in which system each base station in said plurality utilizes time information from one of a number of synchronization sources in order to time said down link, up link and guard periods properly, the method (1000, 1010) being characterized in that it comprises letting each of the base stations in said plurality use information on the inaccuracy of the synchronization source of the base station in order to adapt its downlink periods so that they do not interfere with up link periods in one or more of the other base stations in said plurality.

2. The method (1000, 1020) of claim 1 , according to which the down link periods are adapted by shortening them and shifting them in time by delaying their start.

3. The method (1000, 1030) of claim 2, according to which the down link periods are shortened by silencing one or more symbols at the end of the down link period.

4. The method (1000, 1040) of claim 2, according to which the down link periods are shortened by silencing one or more symbols at the start of the down link period.

5. The method (1000) of any of claims 1-4, according to which the synchronization source is one of the following:

• A satellite based system such as the GPS system,

• Synchronization signals from a transport network to which the system is connected.

• Synchronization signals over the air interface,

• An internal clock in the base station.

6. The method (1000) of any of the previous claims, further comprising the step (1050) of letting one or more of the RBSs in the system shorten its up link period, in order to avoid interference.

7. The method (1000, 1050) of claim 6, according to which the up link period is shortened at its end, by means of silencing one or more symbols during the last part of the up link period.

8. The method (1000) of any of claims 1-5, further comprising the step of letting one or more of the RBSs in said plurality initiate its up link period earlier or later in time than the nominal starting point of the up link period, also comprising the step of letting those RBSs that initiate their up link period earlier or later inform users in the cells of those RBSs of the new starting time of the up link period.

9. The method (1000) of claim 8, according to which the up link period is initiated earlier or later in time based on measurements of a performance indicator representing the interference level, said measurements being carried out in the time interval between two consecutive down link periods.

10. The method (1000) of any of the previous claims, according to which an RBS informs users in its cell of changes in the duration and/or starting point of the downlink periods.

11. A radio base station, RBS₁ (112) for use in a cellular access system (200), for the control of the traffic to and from at least one cell (110, 120, 130, 140) in the system, in which cell the communication between users in the cell and the RBS (112) is divided into an up link period (UL) during which users may transmit to the RBS, and a down link period (DL) during which the RBS may transmit to users in the cell of the RBS, with guard periods being inserted between uplink and downlink periods, the RBS being equipped with means for utilizing time information from one of a number of synchronization sources in order to time said down link, up link and guard periods properly, the RBS (112) being characterized in that it comprises means for using information on the inaccuracy of its synchronization source in order to adapt its downlink periods so that they do not interfere with up link periods in other base stations in the system.

12. The RBS (112) of claim 11 , which is equipped with means for adapting its down link periods by shortening them and shifting them in time by delaying their start.

13. The RBS of (112) of claim 12. which is equipped with means for shortening its down link periods by silencing one or more symbols at the end of its down link periods.

14. The RBS of (112) of claim 12, which is equipped with means for shortening its down link periods by silencing one or more symbols at the start of its down link periods.

15. The RBS (112) of any of claims 11-14, which uses one of the following synchronization sources: • A satellite based system such as the GPS system,

• Synchronization signals from a transport network to which the system is connected, • Synchronization signals over the air interface of the base stations,

• An internal clock in the base station.

16. The RBS (112) of any of claims 11-15, further comprising means for shortening its up link period, in order to avoid interference.

17. The RBS (112) of claim 16, in which the up link period is shortened at its end by means of silencing one or more symbols during the last part of the up link period.

18. The RBS (112) of any of claims 11-17, further comprising means for initiating its up link period earlier or later in time than the nominal starting point of the up link period, as well as means for informing users in the cell of the RBS of the new starting time of the up link period.

19. The RBS (112) of claim 18, being equipped with means for initiating the up link period earlier or later in time, based on measurements of a performance indicator representing the interference level, said measurements being carried out in the time interval between two consecutive down link periods.

20. The RBS (112) of any of claims 11-19, being equipped with means for informing users in its cell of changes in the duration and/or starting point of the downlink periods.