DESCRIPTION OF PREFERRED EMBODIMENTS

[0039] FIG. 1 is a schematic view of a cellular communication network 12 including a plurality of BSs. In this example, an MS 10 is currently connected to a serving BS 14 . The MS 10 may be either in idle mode or in busy mode, as explained above. Initially, when the MS 10 has started the connection with the serving BS 14 , the MS 10 receives from BS 14 a measurement order with a neighbour list including the broadcast frequencies and identities of a plurality of predetermined neighbouring BSs, of which three are shown, 16 a - 16 c . If the MS 10 is in idle mode, the neighbour list is received over a control channel on a broadcast frequency channel containing various cell information. If the MS 10 is in busy mode, i.e., is handed over from an old BS, not shown, to the new serving BS 14 , the neighbour list may be received over a logical control channel embedded in the traffic channel used.

[0040] In both cases, the MS 10 is thereby ordered to perform link quality measurements on the specified broadcast frequencies during idle periods, and to send reports to the serving BS 14 according to a predetermined schedule. The schedule may be specified in the measurement order or may be pre-programmed in the MS 10 . Alternatively, the MS 10 may as yet be unconnected to any serving BS, like for example when the MS 10 has just been powered on. The MS then scans for broadcast frequencies and measures the strongest frequencies in order to register with a serving BS.

[0041] The link quality may be measured as at least one of: a received signal strength (RSS), a carrier-to-interference power ratio (C/I), a carrier power, a Bit Error Rate (BER) or any other link quality parameter. However, the invention is not limited to any particular measuring method or schedule.

[0042] The MS 10 must also identify the BS for each measurement in order to qualify the measurement, which means that the MS 10 receives at least one burst during a TDMA timeslot and attempts to read or estimate the received burst in order to determine an identity of the BS. The BS identification is typically helped by the MS 10 knowing from the received neighbour list which target BS to expect for each frequency.

[0043] According to one aspect of the invention, the measuring and identifying activities are performed simultaneously for each target BS based on the same received signal. In this context, “simultaneously” means that measuring and identifying are performed more or less in parallel by different functional units in the MS, albeit not necessarily exactly at the same time. If the MS 10 thus fails to identify the expected BS properly, the measurement is disregarded and a new identification attempt is made, which also involves a new simultaneous measurement. In this way, it is ensured that the measurement is really made for the identified BS, and that the measurement is up to date.

[0044] In GSM, a BS identity BSIC is included in bursts over the synchronisation channel SCH. In that case, the BS identification may be made by reading an SCH burst being measured.

[0045] The BS identity is in prior systems transmitted in certain logical channel bursts depending on the system defined protocol. As mentioned above for GSM, the Base Station Identity Code BSIC includes the Network Colour Code NCC and the Base station Colour Code BCC. Furthermore, the broadcast frequency channel normally comprises a plurality of multiplexed logical channels including paging channels PCHs and the earlier mentioned common point-to-multipoint channels BCCH, FCCH and SCH, as well as dedicated point-to-point traffic channels TCHs. In GSM systems, the BSIC is normally included only in the SCH, and is not included in the other logical channels. If the MS is in busy mode, it may therefore take some time before it receives an SCH burst for reading the BSIC.

[0046] Moreover, all normal bursts typically include a training bit sequence, which in the GSM case is 26 bits long. Exceptions from “normal” bursts may be the FCCH burst which only contains a sinus wave for frequency synchronisation, and the SCH burst which contains a longer specific training sequence used for initial TDMA burst synchronisation.

[0047] The training sequence in a normal burst from a serving BS is known by the MS, and is used to facilitate synchronisation and decoding or detection of the burst. Typically, there is a set of known training sequences comprising, e.g., eight different sequences, and the one used in a particular burst is identified by a Training Sequence Code TSC of, e.g., 3 bits. For the common channels in GSM, the TSC is identical to the BCC, and for the other channels, the TSC is communicated in channel assignment messages.

[0048] According to another aspect of the invention, the BS identity is related to the training sequence code in a way that is known by mobile stations. An MS receiving a normal burst may then determine the BS identity, regardless of which logical channel the received burst belongs to, by estimating the training sequence and deriving the training sequence code therefrom. In the simplest case, the BS identity is set to be identical with the training sequence code. However, other relationships are possible.

[0049] FIG. 2 illustrates schematically an exemplary normal burst 20 transmitted in a timeslot of a broadcast frequency channel from a target BS included in the neighbour list received by the MS. The normal burst 20 may belong to any logical channel and includes a bit field 22 with a training sequence, in this case arranged approximately in the middle of the burst. The burst 20 may further include various fields 24 , 26 such as data bit fields, tail bit fields etc.

[0050] According to an embodiment of the invention and with reference to the flow chart in FIG. 3 , the MS receives a radio signal in a first step 300 , including a burst, or at least a part thereof, from a target BS. The MS measures the link quality of the burst 20 in step 302 and simultaneously attempts to estimate the training bit sequence in the bit field 22 in step 304 , based on the same received burst. If it is determined in step 306 that the estimation attempt was successful, the MS is able to determine the BS identity in step 308 by deriving the training sequence code therefrom, as explained above. The measurement made in step 302 is then valid and can be used. However, if the estimation attempt was unsuccessful, the measurement is discarded in step 310 .

[0051] In this way, there will be opportunities for the MS to determine the BS identity from all normal bursts. It should be added here that more than one measured signal may be needed to obtain a measurement average value. Furthermore, the training sequence estimation may involve known methods for detecting signals in noise and interference, such as maximum-likelihood detection algorithms or minimum mean-squared error methods.

[0052] Alternatively, the training sequence may be estimated by using channel estimation according to one aspect of the invention, which will be described below. Channel estimation is then conducted for several candidate training sequences, wherein the training sequence that maximises a certain selection metric is selected. An exemplary selection metric is the signal power estimate.

[0053] In some systems, e.g., using EDGE (Enhanced Data rates for GSM/TDMA136 Evolution) technology, more than one modulation form may be used for BS transmissions, and hence further sets of training sequences. In EDGE for example, both 8PSK modulation and GMSK modulation may be used. In this case, the additional training sequences need also be considered during BS identification. For blind detection of modulation, an estimation attempt is required for each possible modulation form.

[0054] The flow chart in FIG. 4 illustrates an exemplary procedure for determining the link quality for a target BS included in a neighbour list according to another aspect of the invention. In this example, two different modulation forms 1 and 2 may be used by the target BS. In a first step 400 , a signal or burst is received by an MS on a channel frequency to be measured according to the list. In step 402 , the link quality of the received signal is measured with respect to a predetermined parameter, such as RSS.

[0055] In parallel with the measuring step 402 , attempts are made to estimate the training sequence (TS) using modulation 1 in step 404 and using modulation 2 in step 406 . In step 408 , the most probable estimation from modulation 1 or 2 is selected depending on the result. In step 410 , it is determined whether the selected estimation attempt was successful, and if so, the BS identity can be determined in step 412 , thereby qualifying the measurement from step 402 . On the other hand, if the selected estimation attempt was unsuccessful, the measurement made in step 402 is discarded in step 416 .

[0056] In cellular networks, it is a general practice to transmit dummy bursts on broadcast frequency channels in timeslots not occupied by any logical channel in order to maintain continuous transmission. A dummy burst carries no intelligible information and typically, a predetermined bit pattern is transmitted. It is thus not possible to determine any BS identity from such dummy bursts, which may occur quite frequently. According to a further aspect of the invention, dummy bursts transmitted over broadcast frequency channels include a BS identity or a training sequence which is related to the BS identity. This will further increase the possibilities for mobile stations to determine the BS identity.

[0057] According to further aspects of the invention, a channel estimation procedure is used, as explained below, for determining the BS identity from a received signal.

[0058] Channel estimation is a well-known method used for estimating a physical channel from received radio signals during an ongoing connection with a serving BS. Channel estimation is conducted at a receiver in the MS for a given TSC based on the corresponding training sequence. Typically, the purpose of channel estimation is to determine a channel impulse response for facilitating the decoding of control data and user data included in the signal. Other purposes for using channel estimation may be to enhance link quality measurements and training sequence estimations, which can hence be used for BS indentification, according to the present invention.

[0059] A received total signal is the sum of contributions from plural BSs reusing the same frequency. The signal contribution from each BS may be a superposition of various propagation reflections of the signal transmitted by the BS, arriving with different delays at the MS. The receiver experiences effectively the signal sent over a physical channel with various channel taps. This channel and its taps may thus be estimated in the receiver by means of channel estimation. As mentioned above, this can be used for estimating a training sequence. When determining which training sequence has most likely been inserted into the transmitted signal, the training sequences, or the equivalent TSCs, may be mapped to BS identities for determining the BS identity according to the present invention.

[0060] Furthermore, an estimate of the received signal power can be calculated as the sum of the squared absolute channel taps. Interference and/or noise power may also be estimated therefrom, and the carrier-to-interference power ratio C/I can be estimated from the signal power and the interference and/or noise power. There are numerous known techniques for channel estimation available, e.g., the so-called least-squares estimation, which will not be described further since the present invention is not concerned with any particular channel estimation method or equipment.

[0061] In prior known mobile systems, channel estimation is only conducted by MSs for signals of the connection with a serving BS. According to one embodiment of the invention, channel estimation is performed on a received signal of a frequency channel for identifying one or more BSs, e.g., according to the neighbour list. Channel estimation can further be used for measuring the link quality, either together with the training sequence estimation or in a separate calculation step.

[0062] An exemplary embodiment of training sequence estimation comprises determining the channel estimates for a set of pre-determined training sequences, calculating a selection metric, and selecting the training sequence that yields the maximum selection metric. The maximum selection metric indicates that the corresponding training sequence was the one most likely transmitted. The set of pre-determined training sequences may be chosen in accordance with the respective TSCs that are mapped to the BS identities in the neighbour list. An exemplary selection metric is the signal power or the C/I estimate derived from the obtained channel estimate as explained above. For instance, the training sequence leading to the maximum signal power or the maximum C/I estimate may be selected. By mapping the selected training sequence to a BS in the neighbour list, the simultaneously made measurement is thus ascribed to that BS for inclusion in the measurement report.

[0063] The link quality measurement may also be obtained by conducting channel estimation for signals from a BS in the neighbour list. Such measurements may be signal power and C/I estimates, which may also be used as selection metric. The benefit of such measurements is that channel estimation inherently separates the wanted signal of the target BS from unwanted signals of other BSs and noise. The accuracy of the measurements is thereby enhanced compared to previous methods based on RSS.

[0064] According to one aspect of the invention, the received signal includes a complete burst period, thus enabling the MS to read any information embedded therein.

[0065] A mobile network may include BSs that are either mutually synchronised or unsynchronised with respect to downlink TDMA burst transmissions. In the case of a mobile network having unsynchronised BSs, an MS needs to scan complete burst periods for detecting training sequences properly, since the time of start and end, respectively, of a burst from a neighbouring BS is not known by the MS. However, this may sometimes be unsuccessful, given the limited time available for receiving other signals than those of a serving BS when the MS is in busy mode. Therefore, the MS may receive only a part of a burst from the neighbouring BS, which may not contain the entire training sequence, depending on an estimation window size. Moreover, the performance of training sequence estimation decreases with increasing estimation window.

[0066] According to a further aspect of the invention, the MS may determine the time of the burst period from the target BS by first searching for a burst synchronisation channel, SCH in the GSM example, on the broadcast frequency. A burst synchronisation channel can be detected and decoded reliably since it typically includes a longer training sequence than for normal bursts. From this logical channel, the MS can obtain timing information of burst periods from this particular BS. This timing information, e.g., in the form of an estimated time offset in relation to the serving BS, may then be used for measuring and decoding of normal bursts, and may also be stored in the MS for later measurements on this frequency channel.

[0067] In the case that bursts from a plurality of unsynchronised BSs reusing the same frequency channel are included in the received signal, the simultaneous measuring and BS identification can be performed by the MS from the different received signal parts coming from the target BSs, constituting the sum of signals simultaneously received from those target BSs on the same frequency. The received total signal can then be evaluated sequentially with respect to each of the target BSs, one at a time, preferably using channel estimation, by detecting the respective training sequences.

[0068] On the other hand, if the target BS is burst synchronised with the serving BS and the MS, it will be much easier for the MS to read a normal burst therefrom and estimate the training sequence properly, since the MS will automatically know where it occurs within the burst in relation to its own synchronisation with the serving BS. In a network with synchronised BSs, the bursts and the training sequences therein will thus arrive to the MS approximately at the same time from plural surrounding BSs, at least within relatively small distances.

[0069] In the case that bursts from a plurality of synchronised BSs reusing the same frequency channel are included in the received signal, the simultaneous measuring and BS identification can be performed by the MS from a part of the received signal that is the sum of signals simultaneously received from those target BSs on the same frequency. The received total signal can then be evaluated sequentially with respect to each of the target BSs, one at a time, preferably using channel estimation, by detecting the respective training sequences.

[0070] Alternatively, it is also possible to perform the evaluation with respect to each BS jointly for plural BSs, reusing the same frequency, on the same received total signal using channel estimation. In this case, the presence of overlaid synchronised training sequences of the BSs is utilised in a joint training sequence estimation procedure that estimates a set of transmitted training sequences, e.g., according to a maximum-likelihood or minimum mean-squared error criterion. A joint channel estimation that estimates the radio channels simultaneously with respect to all target BS signals may thus be used for this purpose. This will of course involve increased complexity, but since plural BSs are measured and identified simultaneously, the processing speed may be increased, and the measurement accuracy may be improved considerably.

[0071] This is illustrated in FIG. 5 as a schematic work flow in an MS 500 receiving a total signal 502 . The signal 502 is the sum of synchronised bursts a-c from three BSs transmitting on the same frequency. In this simplified example, further signal contributions from other BSs are not considered. The total signal 502 enters a channel estimation unit, represented by block 504 . The channel estimation produces three channel estimates 506 corresponding to the three signal contributions from bursts a-c, respectively, by estimating the channels either sequentially or jointly. The three channel estimates 506 are then evaluated, i.e., measured and identified, in parallel, represented by blocks 508 . It will be understood that the measuring and identifying procedure may also involve the received total signal 502 and further calculations in the channel estimation unit 504 . If the respective BS identification was successful, the corresponding results are entered into a measurement report 510 which is finally transmitted to a serving BS 512 . Alternatively, the results can be used by the MS to select a suitable BS to register with in the case of the MS having no serving BS.

[0072] The various embodiments described above have the purpose of increasing the accuracy and relevance of MS measurements on target BSs. The measurements can be tied to specific BSs more reliably, and are at the same time more up to date when reported and/or evaluated than in prior solutions. Furthermore, the inventive link quality estimation is more insensitive to interference, which allows for a tighter reuse of frequencies, in particular broadcast frequencies, in different cells of a mobile network. The tighter frequency reuse thereby enhances the overall traffic capacity in the network.

[0073] In a measurement order, the MS may be directed by a serving BS to select any of the various measuring and identifying schemes described above, all or some of which may be pre-programmed in the MS.

[0074] It has been assumed in the description above that the reported measurement results are primarily used for selecting the most suitable BS for connection with the MS, either just powered on, or in busy or idle mode. However, the measurement results can also be used for estimation of cell relations in order to support network planning. Cell relations include, e.g., an estimated level of interference if the cells are allocated the same or adjacent frequency channels for transmissions. Typical network planning tasks that may be performed based on estimated cell relations include: setting cell patterns and transmission power levels, making antenna adjustments and setting frequency allocation parameters and handover thresholds.

[0075] The invention may be implemented in a computer program loadable into the internal memory of a computer in the mobile station ( 10 ), or in a computer program product stored on a computer usable medium for controlling the mobile station ( 10 ).

[0076] While the invention has been described with reference to specific exemplary embodiments, the description is only intended to illustrate the inventive concept and should not be taken as limiting the scope of the invention. Various alternatives, modifications and equivalents may be used without departing from the spirit of the invention, which is defined by the appended claims.