The present invention relates to a method and related mobile device for cell searching with low memory requirement. The present invention specifically relates to a method and related mobile device for cell searching with low memory requirement in a Long Term Evolution (LTE) system.
Cell search/detection is an important procedure for a mobile device deployed in a cellular network. Typically, reference signals are used to identify a cell which needs to be detected by the mobile device during operation. Reference signals are data symbols which are known at the receiver and are used for parameter estimation tasks, e.g. cell-identity (cell-ID) detection.
In a cellular network, a supplied area is split into cells, each cell being equipped with a base station which serves the mobile stations in that cell. Each cell needs to be uniquely identified if a mobile devices wishes to connect to a cell or if the mobile device is already connected to a cell and wishes to connect to another cell. For this purpose a base station transmits a cell identity (cell-ID) within the Primary Synchronization Signal (PSS) and Secondary Synchronization Signal (SSS). There are 504 unique physical layer cell identities in LTE, grouped into 168 groups of three identities. Three PSS sequences are used to indicate the cell identity within the group and 168 SSS sequences are used to indicate the identity of the group.
In the mobile device, in order to enable reliable data reception, a number of parameter estimation tasks need to be performed, e.g., time synchronization estimation, frequency synchronization estimation, channel estimation, interference level estimation, Doppler spread estimation, power delay profile estimation, feedback information estimation. PSS detection is used for slot timing detection and physical layer ID detection. SSS detection is used for radio frame detection, cyclic prefix (CP) length detection and TDD/FDD detection. SSS detection is based on coherent demodulation in frequency domain.
Present solutions employ oversampling during reference signal detection, for example 2× oversampling during PSS detection and SSS detection in order to achieve good cell search sensitivity.
There are some obvious draw backs associated with 2× oversampling rate processing. With 2× data to be processed, more dynamic power is consumed at silicon side. Moreover, much more hardware resources are required with 2× data, leading to more hardware cost and more leakage power. This is especially true for memory requirement, which usually is the dominant part for hardware cost in cell searcher design.
Hence, what is needed is a solution for cell searching having low memory requirement.
This solution is achieved with a method according to the independent method claim and a related mobile device according to the independent apparatus claim. The dependent claims relate to further aspects of the present solution.
The method and related mobile device according to the invention are described in more detail herein below by way of exemplary embodiments and with reference to the attached drawings, in which:
The present solution relates to a method for searching for a cell in a cellular mobile communication system, the method comprising: receiving in a receiver a signal carrying a reference signal having a first data rate, decimating the signal to generate a down-sampled signal having a second data rate, compensating a phase distortion introduced by the decimating, and correlating the down-sampled signal against a phase-compensated local reference signal or a phase-compensated down-sampled signal against a local reference signal. The second data rate is lower than the first data rate. The second data rate may be a lowest sample rate to obtain a pre-defined accuracy of the method, e.g. 1× oversampling. Phase compensation is necessary due to a phase distortion introduced by a low pass filter associated with decimation in order to fulfil the sampling rate theorem. Compensating a phase distortion may comprise pre-compensating, i.e. offline pre-computing, the reference signal to generate the phase-compensated local reference signal against which the down-sampled signal is correlated against, specifically when using a poly-phase correlator, for example a two phase correlator.
Alternatively, compensating a phase distortion may comprise phase-compensating the down-sampled signal either in time domain or in frequency domain to generate the phase-compensated down-sampled signal used for correlating against the local reference signal being identical to the reference signal.
When buffering the down-sampled signal prior or subsequent to compensating, the buffer size is minimized. With buffering the down-sampled signal, the RF module can be switched off which significantly saves modem power.
For an LTE system, current solutions use 2× sampling data to make sure the PSS peak detection sensitivity is not sacrificed. It was shown that up to 2˜3 dB performance loss could be there if brute force 1× sample rate is used compared to use of 2× sampling rate. With the present solution, however, a poly-phase PSS correlator with pre-compensated local PSS sequences is provided so that almost an equivalent performance can be achieved with 1× sampling.
The time domain signal decimated to 0.96 Ms/s of half-frame p is denoted as y(n,p). During PSS detection, the PSS correlator computes a correlation metric Λr based on sector ID hypothesis r
Λr(2n+u,p)=Σl=0L-1|Σk=0N-1y(n−(l*N+k),p)*su,r*(*N+k)−1)| (1)
where su,r is complex conjugate of time domain pre-compensated PSS sequence with sector ID r (rε0, 1, 2 according to 3GPP spec.) and phase (uε0, 1, 2), L is the number of data segments used for PSS correlation, N is the number of samples per data segment within which coherent combining is used as shown in Eq. 1.
su,r is generated based on 3GPP specified PSS sequence and frequency response of decimator as follows:
sr=IFFT128(Pr*conj(D)) (2)
su,r=sr(2n+u), u=0, 1, n=0 . . . 63 (3)
where Pr is the frequency domain PSS sequence with sector ID r and D is the frequency response of decimator at the frequency range of the PSS sequence. Note that pre-compensation of decimator frequency response is important here since the decimator normally introduces a phase distortion at the edge of a band where PSS sequence is located. With pre-compensation of the local PSS reference sequence, the PSS detector works like a matched filter and thus minimizes the performance loss.
From Λr the final PSS peak detection metric can be generated by accumulating or performing special low cost combining over a number of half-frames. For instance in case accumulating, one or several peaks can be obtained over Pobserv half-frames as shown in Eq. 4.
{tilde over (r)}=argmaxu,r,nΣp=0P
Phase compensation is required before frequency domain SSS detection for two reasons. The first reason is due to the phase distortion introduced by the decimator at the edge of the center 0.96 Mhz band where the PSS sequence is located, i.e. for the same reason as why pre-compensated PSS sequences are required. The second reason is due to the fact that the start position offset of PSS and SSS sequence can be M+0.5 samples at 0.96 Ms/s sample rate, where M can be easily taken care of when data is picked out from the buffer. However, the remaining 0.5 sample effect needs to be compensated ideally for following SSS detection. This can be done after FFT with linear phase compensation over sub-carriers. An aggregate phase, i.e. sum of a linear phase due to the 0.5 sample offset and a pre-compensated phase due to the decimation filter response, can be computed offline and compensation is done once per SSS hypothesis. An alternative to compensate the linear phase due to 0.5 sample offset can be zero-padding in time domain and utilization of a 128 point FFT.
As shown above, the present solution specifically provides a 1× sampling processing in an LTE system for both PSS detection and SSS detection to achieve low cost and low power design. DSP processing techniques are applied in order to make the sensitivity loss due to one times sampling rate processing almost negligible. During PSS detection, 1× sampled data is utilized and a poly-phase correlator with pre-compensated filter coefficients in time domain may be applied. During SSS detection, 1× sampled data is utilized and phase/half sample compensation in frequency domain may be applied.
The present solution offers almost equivalent performance compared to current solutions while it reduces hardware cost and power consumption significantly as will be explained for an offline processing mode and an online processing mode.
Due to the use of 1× sampling rate, the memory buffer requirement is halved. In any cell search implementation, the memory buffer is typically used to buffer the input data during various cell search modes. With a memory buffer present, the RF circuit can be switched off which greatly saves modem power as radio samples can be stored in the memory buffer. The stored radio samples can be utilized during SSS detection.
During offline processing mode, data is buffered into memory for later SSS detection when searcher still operates PSS detection. Apparently, this requires a large buffer to store data. For example, a memory with size of 153.6K bytes is needed in order to store 20 millisecond data, assuming a two times sampling rate, 8 bit precision and 2 receive antennas. The memory requirement can be reduced to 76.8 Kbytes with one times sampling rate as proposed with the present solution. This is very important not only from silicon area point of view, but also from power consumption point of view, since memory leakage power is an important design metric to optimize. With less memory required as proposed in this invention with 1× data sampling, less leakage and less silicon area can be achieved. Real-time processing mode is another example.
In online operations mode, during SSS processing, data is buffered and then picked out for further processing based on peak positions detected at PSS detection step. For continuous operation in order to achieve better searcher sensitivity, a ping pong buffer with size of 10 ms is required. In case of carrier aggregation and simultaneous searching over a number of carriers, the buffer requirements are linearly increased with carrier number. For example in case of 3 carrier aggregation, totally 30 ms data needs to be buffered when searching over 3 carriers simultaneously. Apparently, one time sampling rate buffering with the present solution requires half memories ideas 2× sampling rate (e.g. 115.2 Kbytes versus 230.4 Kbytes).
Not only memory area and its associated power consumption, but also logic area and its associated power consumption can be reduced, since data sample rate is reduced to one time right after decimation, which is a first stage in a search processing chain, as will be shown in detailed description of invention.
The present solution further relates to a mobile device comprising means for receiving a signal carrying a reference signal having a first data rate, means for decimating the signal to generate a down-sampled signal having a second data rate, means for compensating a phase distortion introduced by the decimating, and means for correlating the down-sampled signal against a phase-compensated local reference signal and means for correlating a phase-compensated down-sampled signal against a local reference signal or the down-sampled signal against a phase-compensated local reference signal.
The present solution further relates to a computer-readable medium having computer-executable instructions for performing the method.
Number | Date | Country | Kind |
---|---|---|---|
10 2014 104 090 | Mar 2014 | DE | national |
Number | Name | Date | Kind |
---|---|---|---|
20110007717 | Swarts et al. | Jan 2011 | A1 |
20120122453 | Shin et al. | May 2012 | A1 |
20120281629 | Zhou et al. | Nov 2012 | A1 |
20130176991 | Yi | Jul 2013 | A1 |
20150245308 | Lorca Hernando | Aug 2015 | A1 |
Number | Date | Country | |
---|---|---|---|
20150282055 A1 | Oct 2015 | US |