DE4431047A1 - Separating returning radio signal sections from composite sum signal, for use in e.g. mobile radio network - Google Patents

Abstract

A process for separating high-frequency radio signals from multi-layered composite signals operates on the principle of Fourier transformation. E.g. two signals (1,2) operating within the same predetermined frequency band and sent from two different transmitters (4,5), and are received as a composite signal (3) at a receiver (6). The received signal passes through an analogue/digital converter (7) and I-Q demodulator (8) to digital filters (9). The digital values are then converted with reference to their respective time zones, where the distance (T) between neighbouring time values is used in a Fourier transformation processor, where (T) is greater than the sampling interval. The DFT processor, preferably a special high-speed Fourier transformer based on the Cooley-Tukey-Radix or a Prime-Factor-FFT, changes the digital time zone reading into a spectrum, which is input direct into a data memory, before further processing. The signal sections are accumulated at difference frequency portions in the spectrum, enabling separation of the sections.

Description

A method of separating recurring signal sections a sum signal for sum signals with the properties:

• a) The sum signal is superimposed on a measuring receiver from radio signals C _{i of} one or more transmitters T _i , which are emitted simultaneously within a frequency band to be examined;
• b) some or all of the transmitted radio signals C _{i also} contain recurring signal sequences Q _{i of} any length with the same but otherwise unspecified signal content in the entire signal for the duration of the measurement;
• c) the time interval between two directly successive signal sequences Q _i in the signal transmission is an integer multiple g T of a system time T, g preferably having the value g = 1.

In high frequency technology, it is often necessary to overlay one sum signal formed by individual signals back into these individual signals or at least split parts of it, so as to separate them from each other to carry out separate signals again. That's the way it is for example in cellular networks in which one or more Transmitters carrier signals are transmitted to a receiver and also interfering so-called interferer signals exist at the receiver input superimposed on the actual carrier signal, desirable, both  Information about the carrier signals, as well as about the interferer signals too receive. This includes information about performance ratios, term and Doppler shifts and certain data content that the or Carriers or interferers can identify of interest.

This is not possible with conventional filters or methods, especially when certain partial signals C _{i have} a low power compared to the sum signal and there is no information regarding the absolute transmission times of the signal sections Q _i .

It is therefore an object of the invention to provide a simple method for separating to show special signal parts from the sum signal, which then in can be processed in any way.

This object is achieved by a method according to the main claim solved, advantageous further developments result from subclaims.

The invention is based on the knowledge that digitally modulated signals generally contain repeating signal sections Q _i , the time interval between such sequences being a multiple g of a system time T and g preferably having the value one. These signal sections are often carriers of indirect or direct information such. B. via the transmitter, so quite suitable for problem solving.

According to the invention, the aforementioned signal sections are separated by Fourier transformations. Here, the Fourier transforms are not as applied as a measurement of the power spectral density in a continuous-signal portions, but the _k in the spectra S transformed time ranges Z _k consist of digitized samples of the receiver output signal at an interval of system time T, where T iA is substantially greater than the length of the sampling interval used. The time ranges Z _k to be transformed are determined by their start times t _k . According to the method, the recurring signal sections sought are determined at times t _k as if they had been sent periodically with the system time (ie g = 1 for all transmissions). For this reason, the start times t _k are selected as required for the subsequent further processing of the signal sections Q _i .

In the simplest manner, the first sampling times of the sum signal lying within the measuring period T are taken for the t _k . Other t _k , which can also go beyond the time range of length T, are possible; For example, if slow time A / D converters and memories are used to scan all the time ranges Z _k one after the other and the sampling time T constant during the scanning of the time ranges Z _k during the transition to a new time range by the time Δt is slightly of the order of magnitude of the reciprocal of the signal bandwidth of the sum signal is changed. Because of the periodicity property of the method, the values of the signal sequences Q _i are then obtained at time intervals Δt.

The Fourier transform separates the individual, repeatedly transmitted sequences from the rest of the signal and from one another. Depending on the propagation conditions in the radio channel are obtained, each of these sequences amplified in one or more with respect to a frequency f _m time signals obtained for _m = {y _{1 m,} y _{2 m.} . . } whose values y _{km are taken} from the spectra S _k at a specific frequency point f _m which is the same for all spectra at the times t _{k of} the time signal. The relationship between t _k and S _k arises from the fact that S _k was calculated from the time domain Z _k, which begins at time t _k.

These time signals z _m are then stored for further processing.

If the output signal is available as an IF signal, preferably performed a digital I-Q demodulation, since then the necessary number of samples to be stored less and the number of the Fourier transformations to be performed becomes smaller.

Another way to determine the number of samples to be processed and the The number of necessary Fourier transformations consists in reducing digital pre-filtering of the sampled or digital I-Q demodulated signal.

An advantageous development of the invention results if one instead of the system time T uses a multiple of this time. For example, joins the short training sequence in each transmitted time slot with a GSM signal on. T thus corresponds to the time slot duration, since the time slots are direct  can be sent in succession. Now there is a common case that in a channel only every 8th time slot is sent, which belongs to a certain one Telephone connection heard. In this case, g would always be 8, whichever would adversely affect the separation of the short training sequence. A multiplication of T to 8 times the time slot duration causes here that g = 1 and the separation process described is much more efficient, that is with less measuring and computing effort, can be applied.

The invention is based on schematic drawings of a Embodiment explained in more detail.

Fig. 1 shows the individual transmitters T₁ and T₂, for example, a cellular network to a radio frequency receiver radio signals C₁ and C₂, which are emitted simultaneously within a predetermined frequency band.

Fig. 2 shows the transmitted signals C₁ and C₂ with recurring sequences Q₁ and Q₂.

Fig. 3 shows a possible arrangement for the processing of the received signal into digital samples.

Fig. 4 shows a possible arrangement of the required memory and processor unit for the separation of the signal sections Q _i .

A typical situation of the superposition of two radio signals ( 1 ) and ( 2 ) from different transmitters ( 4 ) and ( 5 ) on the receiver ( 6 ), which can also move at a certain speed, leads to a sum signal ( 3 ) on this receiver. information about the two transmitted signals C₁ and C₂ can be obtained from the separation method described. Such information is contained in the signal sections Q₁ and Q₂. The signal section Q 1 in the signal C 1 is repeated on the transmitter side with a system time T. The signal section Q 2 in the signal C 2 is repeated with the system time T or an integer multiple, as shown in FIG. 2 for g = 2.

The received signal is received and demodulated with an RF receiver ( 6 ). The IF output signal is digitized with the A / D converter ( 7 ) and IQ-demodulated with the down converter ( 8 ) and digitally filtered with the filters ( 9 ).

The complex digital samples obtained in this way are written into a memory ( 10 ) so that they are available as time ranges Z _i for a discrete Fourier transformation. During the system time T, n digital two-valued complex samples of the sum signal are to be delivered from the receiving unit in FIG. 3. Ie the sampling rate is n / d. Then n time ranges Z₁ to Z _{n are} formed, which begin with the first n samples at times t₁ to t _n and then contain values at a distance T. The memory ( 10 ) shown in FIG. 4 is thus filled in columns and the individual time ranges are the n rows of the matrix-shaped memory arrangement. The number N of the columns read and thus the total time t _{g of} the measurement t _g = NT depends on the stationarity of the radio channel. The separation of the signal sections Q _i improves with increasing t _g , as long as the radio channel can be regarded as stationary for this time with respect to the existing propagation paths. Advantageously, N will be chosen so that a fast Fourier transform can be carried out, e.g. B. a Cooley-Tukey-Radix 2 FFT (N = 16, 32, 64, 128,...) Or a Prime Factor FFT (N = 35, 55, 72,...) As described in "DFT / FFT and Convolutional Algorithms ", CS Burrus and TWParks by John Wiley & Sons 1976 . The discrete Fourier transformation of the time ranges Z _i are carried out with the processor ( 11 ), which can be, for example, a special FFT processor, as soon as these have been read in completely. The spectra S _i obtained after the transformations are written directly into the memory ( 12 ) by the processor ( 11 ). In principle, the memory ( 12 ) can be identical to the memory ( 10 ), so that the spectra are written back directly to the memory locations of the time ranges.

The memory ( 12 ) shown in FIG. 4 also has a matrix-like arrangement in whose rows the spectra S _{k are} with respect to the time ranges Z _k with the starting times t _k . The columns then show the time functions z _m which belong to the frequencies f _m and which contain the amplified values of the recurring signal sections Q _i at the times t _k in the values y _km .

The separation of the signal sections Q _i from the signal residue is based on the fact that the data signal components are distributed almost randomly over the entire spectrum during the Fourier transformation. With the period T periodic signal parts in the received sum signal are accumulated f₁ = 0 in the spectral component and thus amplified compared to the rest of the signal. However, since the sequences Q _i are received with a specific frequency deviation Δf, which either originated from frequency deposits in the transmitters T _i or can be traced back to Doppler and fading frequencies, the signals Q _i are accumulated at the frequencies f _{m of} the spectrum if f _m or the frequency f _m shifted by a multiple of the bandwidth B of the spectra S _i coincides with the respective frequency deviation Δf. Thus, different signal sections Q _{i are} normally accumulated at different frequencies f _{m of} the spectra S _i , which makes it possible to separate the Q _{i from} one another.

Claims (3)

1. A method for separating recurring signal sections from a sum signal for sum signals with the properties:

• a) The sum signal is superimposed on a measuring receiver from radio signals C _{i of} one or more transmitters T _i , which are emitted simultaneously within a frequency band to be examined;
• b) some or all of the transmitted radio signals C _{i also} contain recurring signal sequences Q _{i of} any length with the same but otherwise unspecified signal content in the entire signal for the duration of the measurement;
• c) the time interval between two directly successive signal sequences Q _i in the signal transmission is an integer multiple g T of a system time T, g preferably having the value g = 1

with the following known features:

• - The sum signal is received with an RF receiver;
• - The analog receiver output signal is converted into a suitable sequence of digitized real or complex samples within the measurement time; and the following new features
• - Time ranges Z _k for a discrete Fourier transformation are formed from the sample values over a plurality of system times T, by writing sample values into a memory for each time range Z _k at a time interval of the system time T starting with an initial time t _k , the initial times t _k not having the Transmission times of the signal sections Q _i must be synchronized.
• - The time ranges Z _k are converted into spectra S _k by discrete Fourier transformation.
• - For further processing, time signals z _m at the times t _k with respect to the frequencies f _m are provided by storing a numerical value for each value y _{km of} the time signal z _m at time t _k, which value from the spectrum S _k at the frequency point f _m comes from.

2. Procedure according to the main claim with the new feature

• - Instead of the system time T, an integer multiple of this system time is used.