DE102021114327B3 - Method of transmitting signals between multiple transmitters and multiple receivers in a wireless communications network - Google Patents

Abstract

In the method for transmitting signals with a carrier frequency that contain individual data packets, a received signal potentially containing a data packet is shifted in terms of its frequency by a number of frequency offsets in order to feed each correspondingly shifted received signal to an identifier determination unit, which generates an output signal whose magnitude reflects the degree of recognition of the Identifier of the data packet represented. The output signals of the identifier determination units are examined for the output signal with the greatest absolute value, in order then to further process this output signal for decoding the data packet if the absolute value of the output signal exceeds a predetermined threshold value. The identifier determination units are, for example, each swiveled correlators, as a result of which a bank of swiveled correlators is created and used according to the invention.

Description

The invention relates to a method for transmitting signals with a carrier frequency between a plurality of transmitters and a plurality of receivers in a wireless communication network, within which signals containing individual data packets are sent from the transmitters to the receivers in a time-uncoordinated manner, with each data packet being known to the receivers Has identifier and the signals may not have carrier frequencies known to the receivers.

Consider a situation where multiple terminals are accessing a common wireless communication channel in an uncoordinated manner, leading to possible collisions of the signals transmitted by the different terminals. The terminals send packets of data, possibly of any length. Each packet contains a known sequence or synchronization marker (hereinafter also referred to as an identifier) at a predetermined position. This sequence is known to the terminals and the receivers, i. H. it is part of the data transmission protocol. When this sequence is placed at the beginning of the packet, it is usually referred to as a preamble sequence, although it can also be placed at another predefined position at the end of the packet, for example; in this case it is sometimes referred to as a postamble.

The signals sent from the multiple terminals (hereinafter also referred to as transmitters) are received by one or more receivers. Since the terminals' transmissions are not coordinated, the receiver (or receivers) does not know a priori whether the received signal contains the data transmission of one of the users. In order to determine whether transmissions are present, the receiver uses the synchronization marker, which, as already mentioned, is known a priori.

This problem is also sometimes referred to as a frame synchronization problem, user detection problem, or packet search problem.

A commonly used technique to deal with this problem is to use a correlator at the receiver to detect the presence of the synchronization markers in the received signal.

U.S. 2019/0268036 A1 describes a receiver with a series of parallel correlators for fast and efficient synchronization of a received signal which has been subjected to code spreading and/or frequency hopping at the transmitter end and may contain an unknown frequency offset.

wobei der Operator * die (diskrete) Faltung bezeichnet. Das Filter g_k wird typischerweise so gewählt, dass es mit dem Sendefilter der Terminals p_k übereinstimmt, d. h. g_k = p*_-k , wobei der Operator ()* die komplex Konjugierte bezeichnet (diese Wahl maximiert das Signal-Rausch-Verhältnis am Empfänger). In der 1 sei das Sendefilter als analog mit einer Impulsantwort p(t) angenommen, wobei p_k das äquivalente digitale Filter ist.The block diagram of a correlator is in 1 shown. The time-continuous received signal is denoted by y(t). The receiver normally works as follows. First, the receiver samples the received signal and quantizes it to obtain a discrete signal y _k (for the sake of simplicity, no filtering of y(t) is considered, but it is obvious to a person skilled in the art that this is the case in a real receiver ). This signal is then filtered using a receive matched filter g _k to obtain the following $${\mathrm{e.g}}_k=y_k\ast G_k$$

where the * operator denotes the (discrete) convolution. The filter g _k is typically chosen to match the transmit filter of the terminals p _k , i.e. g _k = p* _-k , where the operator ()* denotes the complex conjugate (this choice maximizes the signal-to-noise ratio am Recipient). In the 1 assume the transmit filter is analog with an impulse response p(t), where _pk is the equivalent digital filter.

Here it is assumed that a digital filter g _k is used after sampling and quantization. However, it is obvious to a person skilled in the field of signal processing that the receive filter can also be implemented in analog.

Schließlich wird das Signal w_k mit einer sogenannten Erkennungsschwelle λ verglichen. Liegt das Signal unterhalb der Schwelle, wird angenommen, dass der Synchronisationsmarker im Empfangssignal bei Abtastung k nicht vorhanden ist. Andernfalls wird angenommen, dass der Synchronisationsmarker im Empfangssignal vorhanden ist. Im letzteren Fall, in dem das Vorhandensein eines Pakets erkannt wurde, versucht der Empfänger als nächstes das Paket zu dekodieren. Dazu muss der Empfänger das Signal z_k gespeichert haben, das dann weiterverarbeitet wird, um das Paket zu dekodieren (z. B. wird zuerst das Timing geschätzt, dann der Fading-Koeffizient und schließlich werden Demodulation und Kanaldekodierung durchgeführt).Then the receiver calculates a (cross)-correlation of z _k with the synchronization marker c _k , with k=1,2,...,L. This reduces to filtering z _k with a finite impulse response (FIR) filter whose impulse response is given by h _k = c* _{-k} to obtain a signal denoted w _k , $$w_k={\mathrm{e.g}}_k\ast c_{-k}^{\ast }$$

Finally, the signal w _k is compared with a so-called detection threshold λ. If the signal is below the threshold, it is assumed that the synchronization marker is not present in the received signal at sample k. Otherwise it is assumed that the synchronization marker is present in the received signal. In the latter case, once the presence of a packet has been detected, the receiver next attempts to decode the packet. To do this, the receiver must have stored the signal z _k , which is then further processed to decode the packet (e.g., timing is estimated first, then fading coefficient, and finally demodulation and channel decoding are performed).

Although this approach is not optimal (see [5]), it is not very complex and delivers good results in practice. For this reason, this approach is used in the vast majority of commercial wireless communication systems. However, performance degrades quickly when a carrier frequency offset is present. This carrier frequency offset can be caused either by Doppler shift when the transmitter and receiver move relative to each other, by the instability of oscillators at the transmitter or receiver, or possibly by the fact that the transmitters randomize their transmit frequency.

Other metrics can also be used instead of a correlator, as in 2 shown. For example, the metrics described in [3] and [4] can be used, which are particularly suitable when there is a relatively small carrier offset. It should be mentioned that this, ie a frequency offset due to Doppler shifts, does not have to be the case in the return channel of low earth orbit satellite systems.

Instead, in the presence of a large frequency offset, a more robust alternative is to use a bank of correlators centered around different frequencies (see 3 ). However, this is associated with a high level of complexity at the receiver.

A second alternative is the use of a so-called “swiveled” correlator [see 1] and [2], which uses a fast Fourier transform (FFT) and is less complex. To illustrate the working principle of this approach, recall that the synchronization word (e.g. preamble or postamble) is denoted c _k , with k=1,2,...,L. The sync word is divided into M segments of length L/M, for simplicity it is assumed that L is divided by M, and the ith sync word fragment is denoted by c _i,k , with k = 1,2, .. .,L/M and i = 1,2,...,M. This process is in 4 shown.

wobei der Operator * die diskrete Faltung bezeichnet. Das Filter g_k wird üblicherweise so gewählt, dass es mit dem Sendefilter der Terminals p_k übereinstimmt, d. h. g_k = p*_-k , wobei der Operator ()* die komplex Konjugierte bezeichnet.The swiveled correlator works in such a way that it first samples the received signal y(t) in order to obtain y _k . This signal is then filtered with a matched reception filter (matched filter or optimal filter) g _k to obtain a digital signal z _k , $${\mathrm{e.g}}_k=y_k\ast G_k$$

where the * operator denotes discrete convolution. The filter g _k is usually chosen to match the transmit filter of the terminals p _k , ie g _k = p* _-k , where the operator ()* denotes the complex conjugate.

It is assumed here that a digital reception filter is used. However, it is obvious to a person skilled in the field of signal processing that the filtering can also be carried out in analog with subsequent sampling and quantization.

wobei i = 1,2, ...,M ist, da M Segmente vorliegen.Next, the signal z _k is correlated with all synchronization word segments (e.g. preamble). The result of the correlation of z _k with the i-th synchronization word segment is denoted by w _i,k $$w_{i,k}={\mathrm{e.g}}_k\ast c_{i,k}$$

where i = 1,2,...,M since there are M segments.

Abtastungen auf w_i,k an. Somit berechnet der Empfänger $$a_{i,k}=w_{i,k-\frac{\left(M-i\right)L}{M}}$$

After that, the receiver applies a delay of $\frac{\left(M-i\right)L}{M}$

samples on w _i,k on. Thus, the receiver calculates $$a_{i,k}=w_{i,k-\frac{\left(M-i\right)L}{M}}$$

mit n = 1,2, ...P. Normalerweise wird P > M gewählt, daher wird die FFT mit Zero-Padding berechnet, d. h. man nimmt a_i,k = 0 für i > M an.Next, the receiver computes a P-point FFT of a _i,k over index i. Thus, the receiver calculates $$s_{n,k}={\displaystyle \sum _{i=1}^P{a}_{i,k}{e}^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="DE102021114327B3\_0031.png"\; state="\{\{state\}\}">j</figure-callout>}\frac{2\pi }{P}n\left(i-1\right)}}$$

with n = 1.2,...P. Usually P > M is chosen, so the FFT is calculated with zero-padding, ie one assumes a _i,k = 0 for i > M.

Schließlich wird v_k mit einem Schwellwert λ verglichen. Liegt v_k unterhalb des Schwellwerts, wird angenommen, dass im Empfangssignal bei Abtastung k kein Synchronisationswort und somit auch keine Paketübertragung vorhanden ist. Liegt v_k dagegen über dem Schwellwert, so wird angenommen, dass bei Abtastung k ein Synchronisationswort und damit eine Paketübertragung im Empfangssignal vorhanden ist. In diesem Fall geht der Empfänger zurück zu z_k (das gespeichert sein muss) und versucht, das Paket zu dekodieren, wie zuvor erläutert. Es sei darauf hingewiesen, dass der Empfänger durch Anwendung dieses Verfahrens auch eine Schätzung des Frequenz-Offsets erhält, der das eingehende Paket beeinflusst hat, gegeben durch das Frequenz-Bin, bei dem der Peak v_k, der die Erkennung ausgelöst hat, gefunden wurde. Insbesondere bei der Berechnung von $v_k=\underset{n}{\text{max}}{s}_{n,k}$

liefert der Index n eine Schätzung des Frequenz-Offsets.Next, the receiver calculates $v_k=\underset{n}{\text{Max}}{s}_{n,k}.$

Finally, v _k is compared to a threshold λ. If v _k is below the threshold value, it is assumed that there is no synchronization word and therefore no packet transmission in the received signal during sampling k. On the other hand, if vk is above the threshold value, it is assumed that a synchronization word and thus a packet transmission is present in the received signal during sampling _k . In this case, the receiver goes back to z _k (which must be stored) and tries to decode the packet, as explained before. It should be noted that by applying this method the receiver also gets an estimate of the frequency offset that affected the incoming packet, given the frequency bin at which the peak _vk that triggered detection was found . Especially when calculating $v_k=\underset{n}{\text{Max}}{s}_{n,k}$

love the index n provides an estimate of the frequency offset.

The Swiveled Correlator performs well when the frequency offset is small; however, with a large frequency offset, its performance decreases.

The same can be observed with the techniques proposed in [3] and [4] since they are based on the assumption that the channel only adds Gaussian noise.

After a detailed analysis, it was found that the reason for this drop in performance is none other than the inter-symbol interference (ISI) introduced by the matched receive filter _gk .

In [1] to [4] it is assumed that there is no ISI, but in practice there is always a certain amount of ISI, since the receive filter is not matched to the transmit filter when there is an unknown frequency offset.

It should be noted that situations where there is a large frequency offset (compared to the mean carrier frequency), i. H. when existing schemes severely underperform, are common in practice, and gain importance in many applications. This is the case, for example, when receiving signals sent from ground terminals to LEO satellites. In such situations, the high speed of the satellite with respect to a ground terminal (fixed or mobile) results in very high Doppler shifts.

It is the object of the invention, even if the carrier frequencies of the signals deviate to a greater extent, to still be able to reliably determine the identifier, ie to identify the synchronization marker.

• - empfängerseitig ein potenziell ein Datenpaket enthaltenes Empfangssignal gespeichert wird,
• - das Empfangssignal mit einem ersten Filter gefiltert wird, das einen größeren Frequenzgang als ein auf die Trägerfrequenz abgestimmtes Matched Filter oder Optimalfilter aufweist,
• - dass mittels einer Kennungsbestimmungseinheit des ersten Filters gefilterte Empfangssignal auf die Existenz einer Kennung eines Datenpakets hin untersucht wird,
• - bei erkannter Kennung ein möglicher Offset der Frequenz des Empfangssignals gegenüber der Trägerfrequenz ermittelt wird, und zwar z.B. anhand der Schätzung der Frequenz des gefilterten Empfangssignals mittels eines bekannten Verfahrens aus z.B. der Gruppe der Data-Added-Frequency Schätzverfahren oder durch Verwendung eines Swiveled Korrelators oder einer Korrelatorbank,
• - bei ermitteltem Frequenz-Offset das gespeicherte Empfangssignal hinsichtlich des Frequenz-Offset kompensiert wird und
• - das dergestalt frequenzkompensierte Empfangssignal zum Decodieren des Datenpakets einem zweiten auf die Trägerfrequenz abgestimmten Matched Filter oder Optimalfilter mit sich an dieses anschließender Signalverarbeitungseinheit zugeführt wird.

To solve this problem, according to a first variant of the invention, a method for transmitting signals with a carrier frequency between a plurality of transmitters and a plurality of receivers of a wireless communication network is proposed, within which signals containing individual data packets are sent from the transmitters to the receivers in a non-chronologically coordinated manner be, wherein each data packet has an identifier known to the recipients and the signals may have carrier frequencies that are not known to the recipients, wherein in the method

• - a received signal potentially containing a data packet is stored at the receiver end,
• - the received signal is filtered with a first filter that has a greater frequency response than a matched filter or optimal filter tuned to the carrier frequency,
• - that a received signal filtered by an identifier determination unit of the first filter is examined for the existence of an identifier of a data packet,
• - If the identifier is detected, a possible offset of the frequency of the received signal compared to the carrier frequency is determined, for example by estimating the frequency of the filtered received signal using a known method from, for example, the group of data-added-frequency estimation methods or by using a swiveled correlator or a correlator bank,
• - When the frequency offset is determined, the stored received signal is compensated with regard to the frequency offset and
• - The such frequency-compensated received signal for decoding the data packet is fed to a second matched filter or optimal filter tuned to the carrier frequency with a signal processing unit connected to this.

In this first variant of the method according to the invention, a filter is used in addition to a matched filter (hereinafter sometimes also referred to as optimal filter), which has a greater frequency response than the matched filter tuned to the carrier frequency. The advantage of using such a filter to filter the received signal is that the distortions in the filtered received signal are not too great, even if the carrier frequency of the received signal deviates significantly from the "expected" carrier frequency to which the matched filter is tuned, what means that an examination for the existence of an identifier of a data packet in the received signal is still possible. If the identifier is recognized, a possibly given offset of the carrier frequency of the received signal compared to the "expected" carrier frequency can then be determined, for example based on the estimation of the frequency of the filtered received signal using methods known per se from, for example, the group of data-added-frequency estimation methods ( described e.g. in [6]) or by using a swiveled correlator or a correlator bank with correlators assigned to different frequency shifts.

With the help of the determined frequency offset, the previously stored received signal can then be compensated accordingly and in such a way frequency-compensated received signal are then further processed for decoding the data packet by means of the matched filter and a signal processing unit connected to this.

The deviations in the frequency of the received signals from the "expected" carrier frequency can have various causes. For example, due to a Doppler shift, the frequency of the received signal can shift. It is also possible that, for whatever reason, the transmitters transmit with different frequencies.

With regard to the boundary conditions as far as the frequency responses of the matched filter (or the optimal filter) and the broadband filter used according to the invention are concerned, the following considerations apply quite fundamentally.

The matched filter should be tuned exactly to the transmission frequency, otherwise inter-symbol interference (ISI) is to be expected. However, the matched filter does not necessarily have to be narrow-band. The bandwidth of the matched filter depends on the symbol rate Bs, ie on the number of transmitted symbols per time unit, and also on a parameter beta, which is also conventionally referred to as low-off and can assume values between 0 and 1. The bandwidth B of the matched filter is then $$\text{B}=\text{Bs}\ast \left(1+\text{beta}\right).$$

wobei Fd die Frequenzverschiebung oder die mögliche Frequenzabweichung (z.B. durch Doppler-Effekt) in Hz und Bs die Symbolrate ist.In order to avoid excessive inter-symbol interference, the so-called normalized frequency shift is considered to be an essential factor $$\text{Fd}=\text{Fd}/\text{Bs}\text{,}$$

where Fd is the frequency shift or the possible frequency deviation (e.g. due to the Doppler effect) in Hz and Bs is the symbol rate.

If fd < beta, little or no inter-symbol interference can be expected. On the other hand, if fd > beta, pronounced inter-symbol interference must be expected.

With regard to the frequency response of the "broader" first filter used according to the invention, it is expedient that its frequency response is wider by Fd, i.e. wider by approximately the maximum frequency shift to be expected due to Doppler effects or other influences (i.e. approximately +/- up to 5% or up to 10% or up to 20%) than the matched filter frequency response.

• - ein potenziell ein Datenpaket enthaltendes Empfangssignal durch einen Empfänger empfangen wird,
• - das Empfangssignal bezüglich seiner Frequenz um mehrere Frequenzversatze verschoben wird,
• - jedes um einen Frequenzversatz verschobene Empfangssignal in einem Swiveled Korrelator mit einem auf ein gegenüber der Trägerfrequenz um den Frequenzversatz abgestimmten Matched Filter oder Optimalfilter verarbeitet wird, wobei jeder Swiveled Korrelator ein aus Einzelsignalen zusammengesetztes Ausgangssignal erzeugt,
• - die Ausgangssignale der Swiveled Korrelatoren auf das betragsmäßig größte Ausgangssignal hin untersucht werden und
• - das betragsmäßig größte Ausgangssignal zur Decodierung des Datenpakets weiterverarbeitet wird, wenn der Betrag dieses Ausgangssignal einen vorgegebenen Schwellwert überschreitet.

According to a second variant of the invention, this proposes a method for transmitting signals with a carrier frequency between a plurality of transmitters and a plurality of receivers in a wireless communication network, within which signals containing individual data packets are sent from the transmitters to the receivers in a non-chronologically coordinated manner. wherein each data packet has an identifier known to the recipients and the signals can have carrier frequencies that are not known to the recipients, wherein in the method

• - a received signal potentially containing a data packet is received by a receiver,
• - the frequency of the received signal is shifted by several frequency offsets,
• - each received signal shifted by a frequency offset is processed in a swiveled correlator with a matched filter or optimum filter tuned to the carrier frequency by the frequency offset, with each swiveled correlator generating an output signal composed of individual signals,
• - the output signals of the swiveled correlators are examined for the output signal with the greatest absolute value and
• - the output signal with the greatest absolute value is processed further for decoding the data packet if the amount of this output signal exceeds a predetermined threshold value.

In this method according to the invention, a bank of swiveled correlators is used. Such correlators are known in principle in the prior art. The output signals of the swiveled correlators, which are complex signals, are examined with regard to their magnitudes for the output signal with the greatest magnitude. The output signal with the highest absolute value is then processed further for decoding the data packet, with this output signal with the highest absolute value being at least greater than a predetermined threshold value.

In a further advantageous embodiment of the invention, it can be provided that corresponding individual signals from different swiveled correlators are first combined to form intermediate signals and/or combined with one another, e.g. are added in a weighted manner and/or are subjected to an averaging, and that the intermediate signals are based on the intermediate signal with the greatest absolute value are examined, which thus forms the output signal with the greatest magnitude.

Furthermore, it can advantageously be provided that, instead of the matched filter or the optimum filter, a filter is used per swiveled correlator that has a greater frequency response than the matched filter or the optimum filter. This development of the invention is a combination of the two previously discussed variants of the invention. The advantages of using a filter with a larger frequency response than the matched filter per swiveled correlator are the same as described above in connection with the first variant of the invention. The determination of the frequency response of this filter used instead of the matched filter is also the same as described above.

• - ein potenziell ein Datenpaket enthaltendes Empfangssignal durch einen Empfänger empfangen wird,
• - das Empfangssignal bezüglich seiner Frequenz um mehrere Frequenzversatze verschoben wird,
• - jedes um einen Frequenzversatz verschobene Empfangssignal einer Kennungsbestimmungseinheit zugeführt wird, die ein Ausgangssignal erzeugt, dessen Betrag den Grad der Erkennung der Kennung des Empfangssignals repräsentiert, wobei es sich bei der Gesamtheit der Kennungsbestimmungseinheiten nicht um die Einheiten einer Korrelationsbank handelt,
• - die Ausgangssignale der Kennungsbestimmungseinheiten auf das betragsmäßig größte Ausgangssignal hin untersucht werden und
• - das betragsmäßig größte Ausgangssignal zur Decodierung des Datenpakets weiterverarbeitet wird, wenn der Betrag dieses Ausgangssignals einen vorgegebenen Schwellwert überschreitet.

Finally, the invention also proposes another variant, which is a method for transmitting signals with a carrier frequency between a plurality of transmitters and a plurality of receivers of a wireless communication network, within which signals containing individual data packets are transmitted in a non-chronologically uncoordinated manner by the transmitters are sent to the receivers, with each data packet having an identifier known to the receivers and the signals being able to have carrier frequencies which are not known to the receivers, with the method

• - a received signal potentially containing a data packet is received by a receiver,
• - the frequency of the received signal is shifted by several frequency offsets,
• - each received signal shifted by a frequency offset is fed to an identifier determination unit, which generates an output signal whose amount represents the degree of recognition of the identifier of the received signal, the totality of the identifier determination units not being the units of a correlation bank,
• - the output signals of the identifier determination units are examined for the largest output signal in terms of absolute value and
• - The output signal with the highest absolute value is processed further for decoding the data packet if the absolute value of this output signal exceeds a predetermined threshold value.

Various solutions are therefore proposed with the invention in order to improve the performance of the sync word search block (or frame sync block). The solutions are not mutually exclusive and can be combined.

• 1 ein Blockdiagramm einer Synchronisationssuche mittels eines Korrelators,
• 2 ein Blockdiagramm eines generischen Synchronisationswort-Suchalgorithmus (Rahmensychronisationsalgorithmus),
• 3 ein Blockdiagramm einer Synchronisationssuche mittels einer Bank von Korrelatoren, die um verschiedene Frequenzen zentriert sind,
• 4 eine Veranschaulichung, wie das Synchronisationswort in M Fragmente aufgeteilt wird, die der Einfachheit halber als gleich lang angenommen werden,
• 5 ein Blockdiagramm einer Synchronisationswortsuche mit einem Swiveled Korrelator,
• 6 eine Bank von N Swiveled Korrelatoren, kombiniert mit einem maximalen Betrieb,
• 7 eine Bank von N Swiveled Korrelatoren, unter Verwendung von Clustering und Kombination,
• 8 eine Bank von N generischen Synchronisationswort-Suchblöcken und
• 9 Empfängerbetriebskennlinien, 1 - P_CD (Wahrscheinlichkeit für eine korrekte Erkennung) in Abhängigkeit von P_FA (Wahrscheinlichkeit für einen falschen Alarm) für das System mit einer Präambellänge von 16384 Symbolen, $\frac{E_s}{N_0}=-19.3dB,$
  
  einer Symbolfrequenz von 240 ksymb/s und einer Worst-Case-Doppler-Verschiebung von 50 kHz, wobei mehrere Kurven gezeigt sind, und zwar zwei für die parallele Suche mit 5000 und 3000 Zweigen und drei für eine weitere Suche für den Swiveled Korrelator und wobei die ersten Kurven für einen einzelnen Swiveled Korrelator repräsentativ sind und die letzten beiden Kurven bei vier Swiveled Korrelatoren, parallel gefaltet, gedacht sind, und zwar mit einer Segmentlänge von 1 Symbol (M = L) und einer FFT von 16384 Punkten.

The invention is explained in more detail below with reference to the drawing. In detail show:

• 1 a block diagram of a synchronization search using a correlator,
• 2 a block diagram of a generic synchronization word search algorithm (frame synchronization algorithm),
• 3 a block diagram of a synchronization search using a bank of correlators centered around different frequencies,
• 4 an illustration of how the synchronization word is divided into M fragments, which for the sake of simplicity are assumed to be of equal length,
• 5 a block diagram of a synchronization word search with a swiveled correlator,
• 6 a bank of N Swiveled Correlators combined with a maximum operation,
• 7 a bank of N swiveled correlators, using clustering and combination,
• 8th a bank of N generic sync word search blocks and
• 9 Receiver operating characteristics, 1 - P _CD (probability of correct detection) versus P _FA (probability of false alarm) for the system with a preamble length of 16384 symbols, $\frac{E_s}{N_0}=-19.3\mathrm{i.e}B,$
  
  a symbol frequency of 240 ksymb/s and a worst case Doppler shift of 50 kHz, several curves are shown, two for the parallel search with 5000 and 3000 branches and three for a further search for the swiveled correlator and where the first curves are representative of a single swiveled correlator and the last two curves are intended for four swiveled correlators folded in parallel, with a segment length of 1 symbol (M=L) and an FFT of 16384 points.

The first solution (claim 1) relates to the use of any generic frame synchronization algorithm, while according to the second solution (claim 2) an extension of the swiveled correlator is proposed.

First solution

The first solution is to use a different receive filter g _k . The choice g _k = p* _-k mentioned above is justified when there is no frequency offset, since in this case it is optimal. If there is a (large) frequency offset, the choice is no longer justified. If the frequency offset is large, this choice is even far from optimal.

According to the present invention, the receiver applies any frame synchronization algorithm, but with a receive filter that is not matched to the transmit filter. In the event that the presence of the sync word is detected, the receiver goes back to the sequence y _k and not to z _k as is the case with a standard sync word search algorithm.

an und versucht danach, das Paket zu dekodieren. Zwei Merkmale der vorliegenden Erfindung sind also

1. a) dass der Synchronisationswort-Suchblock (Rahmensynchronisation) ein anderes Empfangsfilter verwendet als der Rest der Empfängerverarbeitungskette und
2. b) dass das Empfangssignal vor der Empfangsfilterung gespeichert wird.

Once the presence of a sync word has been detected, the frequency offset in y _k is estimated and compensated for. In the case of the Swiveled Correlator or Correlator Bank, an estimate of the frequency offset can be obtained directly as a by-product and an extra frequency estimation block may not be required. Next, the receiver applies a matched reception filter (matched filter or optimal filter) $G_k=p_{-k}^{\ast }$

and then tries to decode the packet. Two features of the present invention are thus

1. a) that the synchronization word search block (frame synchronization) uses a different receive filter than the rest of the receiver processing chain and
2. b) that the reception signal is stored before reception filtering.

In general, using a filter _gk with a wider frequency response than p* _-k helps to mitigate the effect of ISI at large frequency offsets. One possibility is to design g _k by running an optimization algorithm in such a way that the level of the ISI is kept as constant as possible within the range of possible frequency offsets in the system.

wobei F_d der maximale Trägeroffset in Hz und F_s die Symbolrate in Hz ist.Assume that _pk is a root raised cosine filter, as is the case in most communication systems. The roll-off factor (or the excess bandwidth) of p _k is denoted by β. f _d denotes the normalized maximum carrier frequency offset, which results as follows $$f_{\mathrm{i.e}}=\frac{f_{\mathrm{i.e}}}{f_s}$$

where F _{d is} the maximum carrier offset in Hz and F _s is the symbol rate in Hz.

A sensible choice for g _k is to choose a root-raised cosine filter with a roll-off factor β'> β. One option is to choose β' as (approximately) β' ≈ β + ƒ _d . β' Choosing something larger than previously defined seems to be a good option in practice. It should be noted that this is only one possibility and the present invention covers any embodiment of the filter.

Second solution

The second solution is to use a bank of sync word search blocks. These synchronization search blocks can, but do not have to, be swiveled correlators. An example of using a bank of swiveled correlators is in 6 shown.

• - Für jeden der N Swiveled Korrelatoren:
  • - Bevor das Empfangssignal y_k dem i-ten Swiveled Korrelator zugeführt wird, wird das Signal um eine Frequenz ƒ_i frequenzverschoben.
  • - Anschließend wird ein Empfangsfilter angewendet, das an das Sendefilter angepasst sein kann, $g_k=p_{-k}^{\ast }.$
    
    Alternativ kann auch ein Filter mit einer anderen Impulsantwort verwendet werden, entsprechend der oben beschriebenen ersten Lösung. Das Ergebnis des Filters ergibt z_k.
  • - Anschließend werden Korrelationen mit den M Synchronisationswortfragmenten berechnet, die w_i,k ergeben und das Signal entsprechend verzögern, um a_i,k zu erhalten.
  • - Als nächstes wird eine P-Punkt-FFT von a_i,k berechnet, die s_n,k, n = 1,2, ...P ergibt. Diese Signale werden als die Ausgangssignale des i-ten Korrelators betrachtet.
  • - Als nächstes werden die Ausgangssignale der mehreren Swiveled Korrelatoren miteinander kombiniert, um eine kombinierte Metrik zu erhalten.
  • - Schließlich wird durch den Vergleich der kombinierten Metrik entschieden, ob das Synchronisationswort im Empfangssignal bei Abtastung k vorhanden war oder nicht.

A bank of N Swiveled Correlators works as follows:

• - For each of the N Swiveled Correlators:
  • - Before the received signal y _k is fed to the i-th swiveled correlator, the signal is frequency-shifted by a frequency ƒ _i .
  • - A reception filter is then applied, which can be adapted to the transmission filter, $G_k=p_{-k}^{\ast }.$
    
    Alternatively, a filter with a different impulse response can also be used, corresponding to the first solution described above. The result of the filter is z _k .
  • - Correlations are then calculated with the M synchronization word fragments that give w _i,k and delay the signal accordingly to obtain a _i,k .
  • - Next, a P-point FFT of a _i,k is computed that gives s _n,k , n = 1,2,...P. These signals are considered to be the output signals of the i-th correlator.
  • - Next, the outputs of the multiple Swiveled Correlators are combined together to obtain a combined metric.
  • - Finally, by comparing the combined metric, it is decided whether or not the synchronization word was present in the received signal at sampling k.

There are different ways to combine the output signals of the individual swiveled correlators. Two options are described below.

für jeden Swiveled Korrelator zu berechnen. Es wird nun einen Index i eingeführt, um die Ausgangssignale der verschiedenen Swiveled Korrelatoren zu unterscheiden. Somit bezeichnet v_i,k das Ausgangssignal des i-ten Swiveled Korrelators für die k -te Empfangsabtastung. Die Metriken v_i,_k werden miteinander kombiniert, indem das Maximum $x_k=\underset{i}{\text{max}}{v}_{i,k}$

genommen wird. Schließlich wird x_k mit einem Schwellwert verglichen, um festzustellen, ob das Synchronisationswort vorhanden war oder nicht.A first possibility is the formation of a maximum, as in 6 shown. In particular, the output signals of each individual swiveled correlator s _n,k are used to calculate a metric $v_k=\underset{n}{\text{Max}}{s}_{n,k}$

for each Swiveled Correlator to calculate. An index i is now introduced to distinguish the output signals of the different swiveled correlators. Thus v _i,k denotes the output of the i th swiveled correlator for the k th receive sample. The metrics v _i , _k are combined by the maximum $x_k=\underset{i}{\text{Max}}{v}_{i,k}$

is taken. Finally, x _k is compared to a threshold to determine whether the sync word was present or not.

However, this is not the only way to combine the output signals of the individual swiveled correlators. Another, more general possibility is to use a clustering and combining approach.

1. 1) Clustern der Ausgangssignale s_i,n,k, die zu ähnlichen Frequenzen (oder der gleichen Frequenz) gehören. Es ist zu beachten, dass s_i,n,k der Frequenz $f_i+\frac{n}{P}$
   
   $$
   
   zugeordnet ist. Diese Cluster werden als Frequenz-Cluster bezeichnet.
2. 2) Kombinieren der verschiedenen Ausgangssignale, die zu demselben Cluster gehören. Dies kann z. B. durch die Verwendung einer gewichteten Summe erreicht werden, was im Folgenden erläutert wird. Es wird angenommen, dass es l Signale gibt, die dem i-ten Frequenz-Cluster zugeordnet sind, und dass mit r_i,b,k, b = 1, 2, ... l das b-te Signal des i-ten Frequenz-Clusters für die k-te Abtastung bezeichnet ist. Man kann die Signale kombinieren, indem eine gewichtete Summe berechnet wird $$u_{i,k}={\displaystyle \sum _{b=1}^l{a}_{i,b}{r}_{i,b,k}}$$
   
   wobei das Signal u_i,k das Ausgangssignal des i-ten Frequenz-Clusters für die k-te Abtastung darstellt.
3. 3) Schließlich wird eine kombinierte Metrik für die k-te Abtastung berechnet, indem das Maximum unter den Ausgangssignalen der l Frequenz-Cluster genommen wird, $q_k=\underset{i}{\text{max}}{u}_{i,k}.$
   

The n-th output signal of the i-th swiveled correlator for sampling k is denoted by s _i,n,k , n=1,2,...P, i=1,2,...,N. One can combine these signals in several steps:

1. 1) Clustering the output signals s _i,n,k belonging to similar frequencies (or the same frequency). Note that s _i,n,k is the frequency $f_i+\frac{n}{P}$
   
   $$
   
   assigned. These clusters are called frequency clusters.
2. 2) Combining the different outputs belonging to the same cluster. This can e.g. B. be achieved by using a weighted sum, which is explained below. It is assumed that there are l signals associated with the i-th frequency cluster and that with r _i,b,k , b=1,2,...l the b-th signal of the i-th frequency clusters for the kth sample. One can combine the signals by calculating a weighted sum $${\mathrm{and}}_{i,k}={\displaystyle \sum _{b=1}^l{a}_{i,b}{\mathrm{right}}_{i,b,k}}$$
   
   where the signal u _i,k represents the output signal of the i th frequency cluster for the k th sample.
3. 3) Finally, a combined metric for the kth sample is calculated by taking the maximum among the outputs of the l frequency clusters, $q_k=\underset{i}{\text{Max}}{\mathrm{and}}_{i,k}.$
   

Finally, q _k is compared to a threshold to determine whether the sync word was present or not.

• - Für jeden der N Synchronisationswort-Suchblöcke:
• - Bevor das Empfangssignal y_k dem i-ten Präambel-Suchblock-Korrelator zugeführt wird, wird das Signal um eine Frequenz ƒ_i frequenzverschoben.
• - Anschließend wird ein Empfangsfilter angewendet, das an das Sendefilter angepasst sein kann, $g_k=p_{-k}^{\ast }.$
  
  Alternativ kann auch ein Filter mit einer anderen Impulsantwort verwendet werden, entsprechend der oben beschriebenen ersten Lösung. Das Ergebnis des Filters ergibt z_i,k.
• - Anschließend wird eine Synchronisationswort-Suchmatrix w_i,k berechnet z. B. unter Verwendung der in [3] und [4] beschriebenen Algorithmen.

Note that the preamble search blocks need not be swiveled correlators but can also be algorithms like those described in [3] and [4]. These algorithms, when used in isolation, only provide information about the presence of a sync word, but no estimate of the frequency shift. 8th Figure 12 shows a block diagram of a block of N (generic) sync word algorithms. In particular, the following applies:

• - For each of the N sync word search blocks:
• - Before the received signal y _k is fed to the i-th preamble search block correlator, the signal is frequency-shifted by a frequency ƒ _i .
• - A reception filter is then applied, which can be adapted to the transmission filter, $G_k=p_{-k}^{\ast }.$
  
  Alternatively, a filter with a different impulse response can also be used, corresponding to the first solution described above. The result of the filter is z _i,k .
• - A synchronization word search matrix w _i,k is then calculated z. B. using the algorithms described in [3] and [4].

In particular, w _i,k is the output of the i th synchronization search block for the k th sample.

zu berechnen. Schließlich wird w_k mit einem Schwellwert verglichen, um festzustellen, ob ein Synchronisationswort im Empfangssignal bei Abtastung k vorhanden war. Es ist zu beachten, dass der Empfänger durch Anwendung dieses Verfahrens auch eine Schätzung des Frequenz-Offsets erhält, der das eingehende Paket beeinflusst hat. Insbesondere sei i = i₀ der Index, der w_i,_k maximiert. Die Schätzung der Empfangsfrequenz ist dann ƒ_i0 , die Frequenzverschiebung, die auf den i₀-ten Zweig der Bank von Synchronisationssuchblöcken angewendet wurde.Next, the output signals of each individual synchronization search block w _i,k are used to calculate a metric $w_k=\underset{i}{\text{Max}}{w}_{i,k}$

to calculate. Finally, w _k is compared to a threshold to determine if a sync word was present in the received signal at sample k. Note that by using this method, the receiver also gets an estimate of the frequency offset that affected the incoming packet. In particular, let i = i ₀ be the index that maximizes w _i , _k . The estimate of the reception frequency is then ƒ _{i 0} , the frequency shift applied to the i ₀ th branch of the bank of synchronization search blocks.

Note that if the sync search block is a simple correlator, one obtains a bank of correlators, which is known. According to the invention, other metrics are to be used here than those described in [3] and [4], so that a different algorithm is obtained.

einer Symbolfrequenz von 240 Ksps und einer Worst-Case-Doppler-Verschiebung von 50 kHz. Es sind mehrere Kurven gezeigt, zwei für die parallele Suche mit 5000 und 3000 Zweigen (siehe die Kurven mit der Liniensymbolik in den ersten beiden Zeilen des in dem Diagramm der 9 eingeblendeten Legendenblocks) und drei für Lösungen mit einem Swiveled Korrelator (siehe die Kurven mit der Symbolik der letzten beiden Zeilen des Legendenblocks der 9). Die erste dieser drei Kurven steht für einen einzelnen Swiveled Korrelator mit einer Segmentlänge von 1 Symbol (M = L) und einer FFT von 16384 Punkten, während die beiden anderen dieser drei Kurven von 4 parallelen Swiveled Korrelatoren ausgehen, mit einer Segmentlänge $\frac{M}{L}=4\text{ Symbole}$

und 4096 bzw. 2048 FFT-Punkten. 9 shows the receiver operating characteristics for a system with a preamble length of 16384 symbols, $\frac{E_s}{N_0}=-19.3\text{dB}\text{,}$

a symbol rate of 240 Ksps and a worst case Doppler ver shift of 50 kHz. Several curves are shown, two for the parallel search with 5000 and 3000 branches (see the curves with the line symbology in the first two rows of the diagram in the 9 superimposed legend block) and three for solutions with a swiveled correlator (see the curves with the symbolism of the last two lines of the legend block of the 9 ). The first of these three curves represents a single swiveled correlator with a segment length of 1 symbol (M=L) and an FFT of 16384 points, while the other two of these three curves assume 4 parallel swiveled correlators with a segment length $\frac{M}{L}=4\text{symbols}$

and 4096 or 2048 FFT points.

The bottom two curves of the diagram of 9 show the results of the concept according to the invention (namely that of 6 with four Swiveled Correlators), while the top three curves of the diagram of the 9 are results of prior art concepts.

As can be seen, the performance of the standard Swiveled Correlator (cyan curve) is slightly worse than that of the 3000-branch parallel search. Nevertheless, the complexity of the standard swiveled correlator is about 90 times lower than that of the 5000-branch parallel search. The lower performance is thus compensated by a significantly reduced complexity.

It should also be noted that the performance obtained with this invention exceeds that of the standard swiveled correlator and is close to the performance obtained with the parallel search algorithm. In addition, the complexity of the scheme described in this invention is between 100 and 200 times lower. Therefore, the algorithm described in this invention shows a good complexity-performance trade-off.

The present invention applies to any wireless communication system in which packet capture is required. It is particularly advantageous whenever frequency offsets that cannot be pre-compensated (e.g. due to the complexity) affect the received signal. An example of applications with practical and commercial relevance is machine communication via low-earth orbiting satellites. In this situation, a potentially large number of ground terminals send data packets to be collected by a satellite, often relying on random access protocols. Efficient packet detection at the receiver plays a key role in such systems to achieve good performance, but can be severely hampered by the very large frequency offset caused by the Doppler shift caused by the high-speed motion of the satellite relative to the transmitters . However, the invention can also be applied to terrestrial radio systems.

1. [1] M Sust, R Kaufmann, F Molitor and GA Björnström Rapid acquisition concept for voice activated CDMA communication, Proc. of IEEE Globecom 1990.
2. [2] Spangenberg SM, Scott I, Mclaughlin S, Povey GJR, Cruickshank GM, Grant PM, An FFT-Based Approach for Fast Acquisition in Spread Spectrum Communication Systems, Wireless Personal Communications Vol. 13, pp 27-56, 2000 , Wiley
3. [3] Choi, Z.Y. and Lee, Y.H.. Frame synchronization in the presence of frequency offset. IEEE Transactions on Communications, Vol. 50, Issue 7, Nov, 2002
4. [4] Wuerll, R., Robert, J., Kilian, G., and Heuberger, G, Optimal One-Shot Detection of Preambles with Frequency Offset. Proceedings of the International Symposium on Advanced Networks and Telecommunication Systems (ANTS), Dec. 2018
5. [5] J. Massey, "Optimum Frame Synchronization," in IEEE Transactions on Communications, vol. 20, no. 2, pp. 115-119, April 1972.
6. [6] U Mengali and M Morelli, "Data-aided frequency estimation for burst digital transmission," in IEEE Transactions on Communications, vol. 45, no. 1, pp. 23-25, Jan. 1997.

Claims (5)

Method for transmitting signals with a carrier frequency between a plurality of transmitters and a plurality of receivers in a wireless communication network, within which signals containing individual data packets are sent from the transmitters to the receivers in a non-chronologically coordinated manner, each data packet having an identifier known to the receivers and the Signals may have carrier frequencies unknown to the receivers, with the method - a received signal potentially containing a data packet being stored at the receiver end, - the received signal being filtered with a first filter which has a greater frequency response than a matched filter or optimum filter tuned to the carrier frequency, - that by means of an identification determination unit of the first filter, the filtered received signal is examined for the existence of an identification of a data packet, - if the identification is recognized, a possible offset of the Frequency of the received signal is determined in relation to the carrier frequency, for example based on the estimation of the frequency of the filtered received signal using a known method from, for example, the group of data-added-frequency estimation methods or by using a swiveled correlator or a correlator bank, - with a determined frequency - Offset the stored received signal is compensated with regard to the frequency offset and - the such frequency-compensated received signal for decoding the data packet is fed to a second matched filter or optimum filter tuned to the carrier frequency with a signal processing unit connected to this. Method for transmitting signals with a carrier frequency between a plurality of transmitters and a plurality of receivers in a wireless communication network, within which signals containing individual data packets are sent from the transmitters to the receivers in a non-chronologically coordinated manner, each data packet having an identifier known to the receivers and the Signals the receivers may not have known carrier frequencies, with the method - a received signal potentially containing a data packet is received by a receiver, - the frequency of the received signal is shifted by several frequency offsets, - each received signal shifted by a frequency offset is processed in a swiveled correlator with a matched filter or optimum filter tuned to the carrier frequency by the frequency offset, with each swiveled correlator generating an output signal composed of individual signals, - the output signals of the swiveled correlators are examined for the output signal with the greatest absolute value and - the output signal with the greatest absolute value is processed further for decoding the data packet if the amount of this output signal exceeds a predetermined threshold value. procedure after claim 2 , characterized in that corresponding individual signals from different swiveled correlators are first combined to form intermediate signals and/or combined with one another, e.g. weighted and added and/or subjected to averaging, and in that the intermediate signals are examined for the intermediate signal with the greatest absolute value, which is thus the forms the largest output signal in terms of absolute value. procedure after claim 2 or 3 , characterized in that instead of the matched filter or the optimum filter, a filter is used per swiveled correlator which has a greater frequency response than the matched filter or the optimum filter. Method for transmitting signals with a carrier frequency between a plurality of transmitters and a plurality of receivers in a wireless communication network, within which signals containing individual data packets are sent from the transmitters to the receivers in a non-chronologically coordinated manner, each data packet having an identifier known to the receivers and the Signals the receivers may not have known carrier frequencies, with the method - a received signal potentially containing a data packet is received by a receiver, - the frequency of the received signal is shifted by several frequency offsets, - each received signal shifted by a frequency offset is fed to an identifier determination unit, which generates an output signal whose amount represents the degree of recognition of the identifier of the received signal, the totality of the identifier determination units not being the units of a correlation bank, - the output signals of the identifier determination units are examined for the largest output signal in terms of absolute value and - The output signal with the highest absolute value is processed further for decoding the data packet if the absolute value of this output signal exceeds a predetermined threshold value.

Images