The present application is a divisional application of the chinese patent application having an application number of 01810755.9, an application date of 2001, 4/5, entitled "base station synchronization for wireless communication systems".
Detailed Description
The preferred embodiments of the present invention will now be described in detail with reference to the drawings, wherein like reference numerals refer to like elements throughout.
Fig. 1 shows a simplified wireless spread spectrum Code Division Multiple Access (CDMA) or Time Division Duplex (TDD) communication system 18. The system 18 includes a plurality of node bs 26, 32, 34, a plurality of RNCs 36, 38, … … 40, a plurality of User Equipments (UEs) 20, 22, 24, and a core network 46. The node B26 in the system 18 communicates with associated user equipment 20-24 (UE). The node B26 has one associated with a single base station 30 or multiple base stations 301……30nAssociated single Site Controller (SC). Each base station has an associated geographic area known as a transmission area. It should be clear that even if base station synchronization is disclosed, transmission region synchronization can be achieved using the present invention.
A set of node bs 26, 32, 34 are connected to a Radio Network Controller (RNC) 36. The radio network controller 36 … … 40 is in turn connected to the core network 46. For simplicity of description, the following description will be made with reference to only one node B, but the present invention can be easily applied to a plurality of node bs.
According to a preferred embodiment, the radio network controller 36 maintains base station synchronization between the internal and node bs 26, 32, 34. Referring to fig. 2, the radio network controller 36 may request the base station through its message generator 53301……30nOr the user equipment 20, 22, 24 makes measurements; the measurement results may be received by its measurement receiving component 54; the state estimate it makes based on these measurements can be updated in an optimal way by its synchronization controller 55; a set of states stored in a covariance matrix 57 may also be managed. The stored states are used for synchronization and represent the time error of each base station 30 relative to a reference, the rate of change of each time error, and the transmission delay between the base stations 30.
The rf network controller 36 also manages a set of measurements stored in a database 59, including: the time of arrival of the measured waveform (i.e., the synchronization burst); the difference in arrival time of the transmissions from the two base stations as measured by the user equipment 20; and estimates of state uncertainty and measurement uncertainty. The rf network controller 36 uses advanced filters, such as kalman filters, to estimate parameters that define relative clock drift, such as the exact distance between one component and another, and to improve the parameters. The estimated time offset is used to infer a frequency mismatch between the frequency references corresponding to the base stations and to perform a plausibility check to ensure that occasional, severely inaccurate measurements do not disrupt the procedure.
The radio network controller 36 is for each base station 301……30nA time factor is specified. This time factor is measured by the radio network controller 36 by selecting one base station as the time base reference for the other base stations. All other base stations are assigned a variable time factor that is updated based on measurements and applied calibrations. The time factor may be an integer (e.g., 0 to 10). A lower quality value indicates better accuracy. Alternatively, the quality may be a continuous (floating point) variable. The reference base station (master base station) is preferably permanently assigned a factor: 0. all remaining base stations are assigned variable values which are adjusted in accordance with the reference base station. To illustrate this time factor hierarchical design,fig. 4 shows a master base station, where all slave base stations 1, 2, 3 are assigned time factor values that can be changed according to the master base station. In one embodiment, the time factor of the slave base station 2 is assigned a value that varies from the slave base station 1, and the slave base station 3 is assigned a value that varies only from the slave base station 2.
The normal operating mode of the rf network controller 36 updates the covariance matrix 57 for the states stored in the rf network controller database 59 every predetermined unit of time (e.g., every 5 seconds or a time set by the operator). One element of the covariance matrix 57 is the estimated difference in time error for each base station.
When the time error difference of a base station exceeds a predetermined threshold, the radio network controller 36 sends a message to support the time error update of the base station. The updating can be done by one of three methods: first, the slave base station gets instructions to measure the signal from a neighboring base station 301、302、……30nBase station time of arrival (BSTOA) of the synchronization burst; second, a neighboring base station 30 with a better factor1、302、……30nObtaining instructions to measure a base station arrival time of the transmission of the slave base station; or, third, a user equipment 20 measures the base station and a neighboring base station 301、302、……30nThe base station arrival time of the synchronization burst.
In the first and second methods using base station to base station BSTOA, the arrival time of a transmission from one base station to another is obtained by observation. Referring to fig. 3, a transmission base station 301A known transmission pattern is transmitted at a predetermined time. The transmission mode may be one from the base station 301The pattern passes through an isolator 64 before being transmitted by the antenna 70. Receiving end base station 301The transmitted waveform is detected using its measurement component 60 when the received signal matches a predetermined valueWhen the symbol is asserted, its measurement component outputs a large value. If the receiver and transmitter are in the same region and have precisely synchronized clocks, the output of the measurement component 60 will appear simultaneously with the transmitted waveform. However, clock misalignment and transmission path delay can result in a time difference.
Equation 1 illustrates the transmission path delay:
r/c + x equation 1
R/c is the distance R between the transmitting unit and the receiving unit divided by the speed of light c. x is the device delay. When the base stations are far apart, R/c usually dominates. Radio waves propagate at the speed of light, approximately 1 foot per nanosecond, or 3 x 10 feet per second8And (4) rice. The aim of base station synchronization is to calibrate each base station to within 1-3 microseconds. Thus, when base station distance is on the order of about 1/2 miles (1 kilometer) or more, distance is an important factor. However, for picocells or microcells that are several tens of meters apart, the distance is not important compared to the dominant measurement accuracy x.
Based on the above considerations, when attempting to synchronize base stations that are far away (more than 1 km), knowledge of the distance is important. The exact location becomes unimportant when attempting to synchronize base stations that are about 50 meters away. After the measurement of the time of arrival of the base station is made, the known propagation distance stored in the rf network controller database 59 is subtracted, and the difference is considered to be a misalignment in time (misadjustment) between the two base stations.
A third method measures the relative time difference of arrival (TDOA) observed by a user equipment between two transmissions sent by two different base stations. The user equipment measures and reports the observed time difference of arrival between transmissions from the two base stations. The radio network controller 36 sends a message to the user equipment 20, 22, 24 to measure the time difference of arrival of the two base stations. Upon receiving this message, the user equipment 20, 22, 24 receives the transmissions of both base stations through its antenna 72 and isolator 64 and measures the time difference of arrival with the user equipment measurement receiving component 68 and transmits the measurement results to its associated base station.
If the location of the ue is known (i.e. its distances r1 and r2 from the two base stations are known) and the timing of both base stations is correct, the time difference of arrival is as shown in equation 2.
(r1-r2)/c formula 2
The measurement deviation derived from this value will be referred to as an indicator of time reference misalignment. As known to those skilled in the art, if the distances r1 and r2 are indeed small enough for a picocell base station, then it is not necessary to know their values. The observed arrival time difference can be directly used as a measure of the transmission time difference.
Once a method is selected, the appropriate message is transmitted to a base station 301……30nOr one user equipment 22, 24, 20. If the message is sent to the base station 302Then the base station 302Is informed which adjacent unit to monitor and measure. If a message is sent to the user equipment 22, the user equipment 22 is informed which base station to measure and in addition the user equipment 22 measures its own base station.
Referring back to fig. 2, the rf network controller 36 has stored in its database 59 each base station 301……30nThe distance between them. The radio network controller 36 next checks whether there is a more recent base station 302Neighboring base stations 30 with better time factor1. Once such a neighboring base station 30 is found1There will be a message sent to the neighbouring base station 301For receiving 'out of sync' base stations 302The measurement result of (1). Alternatively, the RF network controller 36 can send a message to the "out of sync" base station 302And requires it to receive the neighboring base station 301The measurement result of (1). Book toolIn the bulk embodiment, the base station subject to the requirement, i.e. "out of sync" base station 302And subsequently receives the "in-sync" base station 301And sends the measurement back to the radio network controller measurement component 54. The radio network controller measurement component 54 forwards the measurement values to the synchronization controller 55, which calculates the transmission time of the measurement result by subtracting the propagation time R/C.
Once the transmission time is calculated by the rnc sync controller 55, this value is compared to the value stored in the rnc database 59. The radio network controller synchronization controller 55 then calculates the kalman filter gain and updates the states in the covariance matrix 57 using the difference between the calculated arrival time and the predetermined arrival time and the common gain. If the difference exceeds a particular threshold, the RF network controller message generator 53 will send another message to the "out of sync" base station 302To adjust its timing or its reference frequency for communication with other base stations 301……30nSynchronization is maintained under the control of the radio network controller 36.
Base station 302Performs the required adjustments and reports them back to the rf network controller measurement component 54. The database in the radio network controller 36 is updated, including with the slave base stations 302Calibration of the time reference, its rate of time variation, updates of its covariance matrix 57 (most importantly including its estimated RMS time error and drift error), and updates of its time factor. Referring to fig. 4, if the time reference of one base station is calibrated based on comparison with other base stations, that base station is never assigned a factor equal to or better than its primary base station. This method ensures stability. For the sake of illustration, if a slave base station 2 is to be calibrated, the value assigned to the slave base station 2 can only be lower than the value of the time factor of the slave base station 1. In doing so, it is ensured that the time factor of a base station does not remain synchronized with a slave base station of the same or lower class, which is the most subordinate base stationEventually, a group of base stations will drift and "lose synchronization" with the master base station.
As disclosed previously, another type of base station 30 that receives measurements to adjust for "out of sync2The method of (2) uses one user equipment 20, 22, 24. If the radio network controller 36 selects this method, it sends a message to the user equipment 22 to measure the "out of sync" base station 302Synchronous burst and "in-sync" base station 301Is synchronized to the burst. Once the user equipment 22 receives the measurements, the measurements are sent to the radio network controller 36 and processed. Similar to the method described above, the measurements are compared to known measurements stored in the RF network controller database 56 and covariance matrix 57 and sent to the "out of sync" base station 302The adjustment measurements of (a) are compared.
A flow chart of the system according to the preferred embodiment is shown in fig. 5a and 5 b. The rf network controller 36 updates the covariance matrix 57 and the database 59 every unit time (step 501). When the radio network controller 36 detects a base station 302……30nExceeds a predetermined threshold (step 502), the radio network controller 36 determines whether to use a base station to measure the base station time of arrival or a user equipment to measure the time difference of arrival to update the "out of sync" base station time error difference (step 503). If the radio network controller 36 decides to measure the base station time of arrival, it sends a message to a base station adjacent to the "out of sync" base station to measure the time of arrival of the base station, or sends a message to the "out of sync" base station to measure the time of arrival of the adjacent base station (step 504). The corresponding base station makes the necessary measurements (step 505) and transmits the measurement results to the radio network controller 36 (step 506). If the radio network controller 36 decides to measure the time difference of arrival, the radio network controller 36 sends a message to a user equipment to measure the time difference of arrival between two base stations (step 507a), one of which is "out of synchronization"The base station of (1). The user equipment measures the time difference of arrival of each base station (step 507b) and transmits the difference of the measurement results to the radio network controller 36 (step 507 c). When the appropriate measurements are received by the rf network controller 36 (step 508), the rf network controller 36 compares the measurements to the values stored in the rf network controller database 59 (step 509). If the difference exceeds a particular threshold, then the radio network controller 36 sends a message to the "out of sync" base station to adjust its timing or its reference frequency based on the difference (step 510). The "out of sync" base station adjusts as required (step 511) and reports the adjustment back to the radio network controller 36 (step 512). The rf network controller database 59 and covariance matrix 57 are then updated to accommodate the new values (step 513).
A preferred embodiment is a system and method residing in each of the radio network controllers 36. In the prior art, a master radio network controller (C-RNC) communicates directly with its base stations, and a serving radio network controller (S-RNC) communicates directly with its user equipment. In case the neighbouring base stations are under control of different radio network controllers, it may be necessary to increase the communication between the master radio network controller (C-RNC) and the serving radio network controller (S-RNC) controlling the neighbouring base stations and the user equipment.
An alternative embodiment requires that each pair of base stations that can listen to each other move their own frequency closer to the other. The relative adjustment is determined by a unique set of weights assigned to each base station and stored in the rnc database 59. The process of adjusting each base station is the same as described above for the preferred embodiment, except that the "in-sync" and "out-of-sync" base stations are adjusted based on the weights assigned to the respective base stations. One can obtain different degrees of centrality between fully centralised and fully distributed according to different weights.
The preferred embodiment enables a radio network controller 36 to align and/or frequency timeThe rate calibration is sent to a base station 301……30n. The master base station is responsible for ensuring that each of its base stations has a time reference dependent from it that remains accurate within a specified limit. The rf network controller 36 assumes in its algorithms and calibrations that there is negligible error between the master base station and its base stations, and therefore assumes that all base stations have the same time reference.
As a result, the rf network controller 36 does not attempt to estimate a single time error between the host base station and its base station, and the host base station must eliminate or compensate for the time error between the host base station and all other base stations because the associated rf network controller 36 is not calibrated. This embodiment presents a neat interface between a radio network controller 36 and a master base station. This embodiment enables the master base station to apply its own solution for slave base station synchronization processes that are well suited to picocell base stations.
In an alternative embodiment, each base station has a separate time and frequency reference that enables a radio network controller 36 to send time alignment and/or frequency alignment to each base station. The rf network controller 36 estimates states representing the time and frequency errors of each base station in its algorithms and calibrations.
As a result, the rf network controller 36 attempts to estimate a single time error between each base station and the master base station, and measurements involving one base station do not help estimate the state of another base station. Thus, the base station manufacturer need only provide loosely-bounded errors with respect to base station timing and time offsets, and each base station must have acceptable radio connectivity with another base station (the same or a different base station).
This alternative embodiment facilitates a wide cellular transmission area where the distance between base stations is large. The ability to calibrate the base station of one slave to the master base station time reference by measurements involving another base station slave to the same master base station is limited.
Each base station in this alternative embodiment uses a separate time reference, but the master base station provides a frequency reference. An rf network controller 36 transmits time alignment for each base station individually and/or transmits an individual frequency alignment to a master base station. The rf network controller 36 ensures that the clock of each base station is slaved in frequency to the clock of the master base station. The radio network controller 36 assumes in its algorithms and calibrations that there is negligible drift error between the master base station and its designated base station, but estimates the offset which is treated as a constant.
As a result, the radio network controller 36 estimates a single time error between the master base station and its base station and a common frequency drift of each base station relative to the master base station.
This alternative embodiment has similar features to the previous one in that base stations that are far from the main base station are beneficial. Such an embodiment provides a mechanism to initiate long distance time mismatches. Based on the assumption that these time offsets are stable, this embodiment utilizes a measurement involving any base station whose frequency is slaved to the master base station clock to update the drift rates of all base stations slaved to the same master base station.
Another alternative embodiment has the rf network controller 36, the rf network controller 36 being able to provide estimates to the master base station to support its synchronization process to the slave base stations. An rf network controller 36 sends the time alignment and/or frequency alignment of each associated base station to its corresponding master base station. The master base station ensures that each of its associated base stations has a time reference that is slaved to itself and that remains accurate within a specified limit. The master base station may choose to use estimates unique to the base station to assist in base station synchronization. The rf network controller 36 creates an optimal estimate of the time and frequency error between the master base station and its base station in its algorithms and calibrations. In performing the state estimation, the radio network controller 36 weights the relative confidence between the measurements and the base station error uncertainty.
As a result, the rf network controller 36 attempts to estimate a single time error between the master base station and its base station, and the master base station eliminates and/or compensates for timing errors between the master base station and each base station that is slaved to its time reference, or requires assistance from the rf network controller 36.
Although the present invention has been described with reference to preferred embodiments, other variants, which are within the scope of the invention as set forth in the appended claims, will be apparent to those skilled in the art.