Electrical computers: arithmetic processing and calculating – Electrical digital calculating computer – Particular function performed
Reexamination Certificate
1998-04-30
2001-02-27
Malzahn, David H. (Department: 2787)
Electrical computers: arithmetic processing and calculating
Electrical digital calculating computer
Particular function performed
C708S312000
Reexamination Certificate
active
06195675
ABSTRACT:
This invention relates to tone detection using discrete Fourier transform (DFT) techniques. The invention is particularly concerned with detection of a supervisory audio tone (SAT) in wireless communications systems such as AMPS (Advanced Mobile Phone System).
BACKGROUND OF THE INVENTION
In a wireless communications system such as AMPS, FM (frequency modulation) is used for communicating voice signals together with a supervisory audio tone (SAT) which is transmitted by a base station with a nominal frequency selected from three possible nominal frequencies of 5970, 6000, and 6030 Hz, and an FM deviation of ±2 kHz±10%. SAT components of the signal transmitted by the base station are transponded by a mobile terminal back to the base station using a phase locked loop (PLL). The base station is required to detect the presence and loss of SAT within attack and release times, respectively, of 200 ms. The SAT is required to be present all the time for the voice channel, and voice calls can be interrupted if a valid SAT is not detected. Accordingly, reliable SAT detection is important for proper operation of the system.
In practice, reliability of SAT detection can be of concern in situations where there is a low C/I (signal carrier to interference) ratio, for example under co-channel interference (CCI) conditions, and situations where the mobile terminal may not fully comply with specifications. For example, in fading and shadowing conditions, the PLL in the mobile terminal may not be able to track well, so that the transponded SAT can drift in frequency. In addition, the FM deviation of the transponded SAT frequency may be different from the specified ±2 kHz±10%, also adversely affecting SAT detection by the base station.
A digital PLL can be used to detect a sinusoidal signal, but is not well suited to SAT detection because it requires a relatively long capture time and does not work well in the fading channel environment of wireless communications systems. A matched filter or coherent detection technique can be used, but this only works well with SAT frequencies that are very accurate. The SAT detection performance of matched filters is seriously degraded for small frequency differences for example of 3 Hz, so that matched filters can not be used for detecting SAT from out-of-specification mobile terminals. Auto-correlation based SAT detectors have been used because they are not sensitive to SAT frequency variation, but they do not provide an optimal SAT detection.
McPherson et al. U.S. Pat. No. 4,698,769, issued Oct. 6, 1987 and entitled “Supervisory Audio Tone Detection In A Radio Channel”, describes a base station SAT detector in which a DFT is applied to complex numbers derived by accumulation from samples of the received signal, and the powers at the SAT frequencies are compared with threshold levels to determine whether or not the respective SAT is present in the received signal. This patent describes the use of an 8-point DFT with an approximation used to simplify the calculations, but in practice a larger and more accurate DFT is required to provide a desired frequency resolution for SAT detection. Typically, a 32-point DFT may be desired. The resources required for computing a 32-point DFT are considerably greater than those required for computing an 8-point DFT.
Block computation of a 32-point DFT in known manner also creates a high peak digital signal processing (DSP) load for computing the DFT periodically, with relatively low DSP loads at other times. In practice in a base station of a wireless communications system it is desirable to avoid such high peak DSP loads.
Accordingly, an object of this invention is to provide an improved method of and apparatus for DFT computation for tone detection, particularly for detecting SAT in a wireless communications system.
SUMMARY OF THE INVENTION
One aspect of this invention provides a method of determining at least one frequency component of a sampled input signal using an N-point DFT, where N=2
M
and M is an integer greater than 1, the DFT comprising M stages of butterfly computations from at least one input for the successive samples of the input signal to a respective output for each frequency component, comprising the steps of: providing each of said stages with up to 2
m−1
butterfly computations where m is a number of the stage from 1 to M from said at least one input to said respective output(s); determining said at least one frequency component at its respective output from the stage M; and for each of the stages from 1 to M−1: determining 1 to 2
m
intermediate results using the respective butterfly computation(s) and storing the determined intermediate result(s); and deriving 1 to 2
m
intermediate results required, in addition to the 1 to 2
m
intermediate results determined in respect of a current sample of the input signal, for butterfly computation(s) in the next subsequent stage from the stored intermediate result(s) determined in respect of preceding sample(s) of the input signal.
Another aspect of this invention provides a method of computing at least one output of a discrete Fourier transform (DFT) from input signal samples using a digital signal processor (DSP) providing a plurality of computation stages and storage locations for storing intermediate results between the stages, comprising the steps of: for at least one of the stages, determining only a subset of the intermediate results required for computation by the next stage; storing the determined subset of intermediate results in respective ones of said storage locations; and producing each other intermediate result required for computation by the next stage by shifting relative to the storage locations a respective stored intermediate result of said subset determined in respect of a preceding input signal sample.
Preferably each of the stages comprises a butterfly computation stage having two inputs and at least one output. Conveniently the DFT is an N-point DFT where N=2
M
and M is an integer greater than 1, there are M stages, each of the stages m from 1 to M−1 determines at most 2
m
intermediate results in the respective subset for a current input signal sample, and each of up to 2
m
other intermediate results required for computation by the next stage is produced by said shifting of a respective stored intermediate result of the respective subset determined in respect of an input signal sample preceding the current input signal sample by 2
M−m−1
samples. The method is used to advantage for computing at least log
2
(N) outputs of the DFT, and can be used for computing N outputs of the DFT.
The invention also provides a method of detecting at least one predetermined frequency component in a signal, comprising the steps of computing, for each predetermined frequency component, at least one respective output corresponding to said frequency component of a DFT by the method recited above, samples of said signal being supplied as the input signal samples to the DFT, and detecting presence or absence of the predetermined frequency component from the respective output of the DFT.
A further aspect of the invention provides a method of detecting a frequency component in a signal, comprising the steps of: supplying samples of the signal to at least one input of a discrete Fourier transform (DFT) having at least one output corresponding to said frequency component; and monitoring a signal at said at least one output of the DFT to detect said frequency component in the signal; the DFT comprising a plurality of sequential computation stages between said at least one input and said at least one output of the DFT, and storage locations for storing intermediate results between successive stages; wherein the DFT computes the signal at said at least one output using intermediate results, for at least one of the computation stages, which are a combination of intermediate results determined in respect of a current sample of the input signal and intermediate results which have been stored in th
Tong Wen
Wang Rui
Malzahn David H.
Nortel Networks Limited
LandOfFree
Tone detection using discrete fourier transform techniques does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Tone detection using discrete fourier transform techniques, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Tone detection using discrete fourier transform techniques will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2593203