CN102624662A - Incoherent detection technology suitable for DMR digital trunking communication system - Google Patents

Abstract

The invention belongs to the field of wireless communication and relates to an incoherent detection technology suitable for a DMR (Digital Mobile Radio) digital trunking communication system. Firstly, defects of both a conventional incoherent detection technology and an existing multiple symbol detection algorithm are analyzed in detail; then according to the maximum likelihood ratio criterion, an incoherent multiple symbol detection algorithm suitable for a 4-CPFSK (Continuous Phase Frequency Shift Keying) signal in a DMR standard frame is disclosed; and simultaneously, a decomposed linear signal replaces an original reference signal to simplify the structure of a receiver. The receiver corresponding to the algorithm disclosed by the invention is formed by three front-end matched filters and a following viterbi operational part and the complexity of the receiver is effectively reduced. When the calculation of the branch metric is carried out, the judgment accuracy is improved by utilizing information of a plurality of pieces of code element time. A computer simulation result in the Matlab environment proves that with the increase of observing windows, the performance of detecting the 4-CPFSK signal of the algorithm disclosed by the invention gradually becomes better.

Description

A kind of incoherent detection technology that is applicable to the DMR digital clustering communication system

Technical field

The invention belongs to wireless communication field, particularly a kind of incoherent detection technology that is applicable to the DMR digital clustering communication system.

Background technology

DMR (Digital Mobile Radio) digital cluster communication standard be ETSI (ETS association) for the low and middle-end specialty that satisfies European various countries and business user to the needs of mobile communication and the open standard that designs, works out.DMR has and efficiently utilizes frequency spectrum resource, and Da Qu makes networking mode, the conventional advantage of compatible simulation.Its business function is abundant, can expand, and back compatible, simultaneity factor and terminal cost are lower, the networking rapid speed, Total Cost Ownership is lower.With respect to ripe TETRA and APCO25 two big digital standard products, have low price, advantage that the technical sophistication degree is low.Domestic manufacturers are all at the product of actively researching and developing the DMR standard at present.

Produce by carrier frequency shift for general fsk signal; The digital information that its reflection will be sent; Belonging to memoryless modulation system, generally adopt the frequency switching way to accomplish modulation, is burst type; Signal phase is discontinuous, so caused signal outside main frequency range, bigger frequency secondary lobe to be arranged.In the DMR standard, to send signal and adopt the 4-CPFSK modulation, message transmission rate is 9.6Kb/s, 4-CPFSK is a kind of Continuous Phase Modulation (CPM) mode of multi-system, can obtain the availability of frequency spectrum efficiently.In modulated process, emission filter is a square root raised cosine filter, the roll-off factor α of filter value between [0,1], and its value has material impact to systematic function.When α was big, the decay of time domain waveform was fast, and it is little to vibrate, and this is favourable for the influence that reduces intersymbol interference and timing error, but system's band occupancy broadens, and band efficiency has reduced, and the while in-band noise also can increase the influence of signal; When α hour, the band system band utilance is improved, The noise has been weakened, but because the influence of intersymbol interference and timing error increases, error rate of system becomes greatly.In the DMR standard, roll-off factor α=0.2 of filter.

Under the white Gaussian noise channel, the 4-CPFSK signal has relevant and incoherent two big types of detection techniques.Coherent detection needs this locality that synchronizing information is provided; Although the simultaneous techniques of binary system continuous phase modulated signal has mature theory; But it is a difficult point that these theories are generalized to multi-system partial response CPM signal always; Because the search of coherent synchronization algorithm is along with the parameter dimension increases and the very difficulty that becomes, it is chronic that carrier synchronization is recovered, and have bigger losing lock probability.Because the 4-CPFSK signal is a kind of multi-system partial response continuous phase modulated signal, the extraction of synchronizing information is difficulty relatively, and coherent detection technology is not suitable for the 4-CPFSK signal.The incoherent detection technology need not by the homophase carrier wave that is difficult to obtain, and hardware implementation cost is low, is fit to be applied in the detection that realizes in the DMR digital cluster system signal.

Traditional incoherent detection technology generally adopts method, for example common Differential Detection and the amplitude limit frequency discrimination of symbol (Symbol-by-Symbol) detection one by one.The 4-CPFSK signal has memory, and this is to introduce through the continuity of phase place, and the method for symbol detection has not made full use of these characteristics of signal one by one, and receiver performance can be poor.Many symbol detection propose for high-performance detects the CPFSK signal at first, also are applicable to the CPM signal more general than CPFSK signal simultaneously.Many symbol detection are meant that receiver receives the signal of a symbol time, do not adjudicate immediately, but receive after the signal of a plurality of symbol times again, this code element are adjudicated, to reduce the error probability of bit decision again.Many symbol detection can obtain the performance more excellent than traditional incoherent technology, watch window be provided with enough big the time, the performance of approaching optimum coherent receiver almost.At present domestic this Study on Technology still is in the starting stage, achievement in research is few.External representative studies achievement has following two:

1. at " IEEE Transactions On Communications "; Vol.com-22; In 1974 deliver among the No.8 " Coherent and Non-coherent Detection of CPFSK ", introduced a kind of non-coherent demodulation technology of CPFSK signal, this technology adopts the method for maximal possibility estimation; On the basis that receives a plurality of symbol time signals, the code element of centre is adjudicated.When watch window is enough big; Can reach the performance of optimum coherent demodulation; But the situation of binary system CPFSK only introduced in article, is not suitable for the quaternary CPFSK signal in the DMR digital cluster system, and the increase that the number of matched filter is accompanied by partial response length and many symbols watch window in should the technology corresponding receiver structure is exponential increase; Even this technology popularization implements and still need design a lot of complicated filter devices to the situation of multi-system.

2. at " IEEE Transactions On Communications "; Vol.41; In 1993 deliver among the No.1 " Maximum Likelihood Block Detection of Non-coherent Continuous Phase Modulation "; Introduced a kind of incoherent many symbol detection technique of CPM signal, this technology adopts the method for maximal possibility estimation, utilizes the reception signal decision of N symbol time to go out N code element information.Receiver performance has very significantly lifting compared with traditional incoherent technology; The increase performance that is accompanied by watch window improves gradually; The number of matched filter also is controlled in the reasonable range in the receiver structure, but should technology only be applicable to the situation of total regression CPM, for general partial response situation; Like the substandard 4-CPFSK signal of DMR, the technological means that needs to seek other solves.

In sum, the detection of the substandard 4-CPFSK signal of the inapplicable DMR of existing many symbol detection technique.

Summary of the invention

Main purpose of the present invention is that the detection of 4-CPFSK signal coherence is difficult to realize under the DMR digital cluster system standard in order to solve; Traditional poor again problem of incoherent detection performance; A kind of incoherent many symbol detection technique that are applicable to the 4-CPFSK signal of this standard are proposed; In algorithm, introduce simultaneously the linear model of multi-system signal, overcome existing many symbol detection receiver complexity is exponential increase along with partial response length and many symbols watch window problem.

The present invention derives the incoherent detection judgement amount of 4-CPFSK signal according to the maximum likelihood ratio criterion, adopts the Mengali algorithm with signal linear expression, draws final many symbol detection judgement amount, designs the receiver structure of simplification.The present invention adopts following steps:

Step 1:, define incoherent many symbol detection judgement amount according to the maximum likelihood ratio criterion.

The maximum likelihood ratio criterion is to select the maximum situation of likelihood function as court verdict.The 4-CPFSK signal is the continuous phase response signal, and the continuity of phase place has caused the Memorability between signal, and in order to improve the correctness of judgement, the present invention once observes the reception signal of a plurality of symbol times.Under additive white Gaussian noise channel (AWGN); Can calculate a plurality of symbol times and receive the corresponding likelihood function of signal; The method of taking simultaneously to make even equal to the phase place in the likelihood function is eliminated the influence of unknown parameter to judgement, then according to maximum-likelihood criterion, chooses incoherent judgement amount.But this judgement amount is based on the complex envelope method of exhaustion of original signal waveform; Along with partial response length, the signal waveform more sophisticated that becomes is carried out matched filtering on this basis to received signal; The number of filter is huge; Matched filter number and Viterbi status number in can't the simplified receiver structure, the present invention utilizes the linear breakdown of 4-CPFSK signal to replace the primary signal of reference, and the back of deriving obtains final incoherent many symbol detection judgement amount.

Step 2: the design of three front end matched filters in many symbol detection receiver.

The present invention adopts the linear model simplified receiver structure of signal, and with the decreased number to three of filter, wherein three of respective signal are linear respectively decomposes real waveforms for matched filter.According to the Mengali algorithm, can perfectly be similar to the substandard 4-CPFSK signal of DMR with the linear superposition of three amplitude modulation(PAM) pulses (AMP), so noncoherent receiver is made up of three front end filter.Filter design procedure is at first with quarternary information { ± 1; ± 3} is expressed as two binary message { ± 1}; Accomplish the linearity of two binary signal components and decompose corresponding amplitude modulation(PAM) impulse waveform; With the real waveform of these decomposition according to certain principle combinations together, the linearity that finally obtains quaternary signal is decomposed real waveform then, the front end matched filter that promptly needs in many symbol detection receiver.

Step 3: baseband signal through matched filter filtering, is obtained sampled value in the time point sampling, calculate the incoherent detection judgement amount, choose the maximum information sequence of judgement amount as the output result.

In the calculating of incoherent many symbol detection judgement amount; Need use baseband signal and carry out the sampled value after the matched filtering through filter, therefore, with the matched filter of baseband receiving signals through designing; And, obtain to calculate in the judgement amount needed numerical information at the peak-peak point sampling.When the situation of carrying out the judgement amount maximum was chosen, along with the increase that detects sequence, detector need be selected a kind of situation of judgement amount maximum from very many possibilities, implement more complicated.Can select { α _N-Nα _N-1As state, a _nAs input, the Viterbi state transition diagram that draws selects for use viterbi algorithm to simplify the process that detects.

Many symbol detection technique that the present invention proposes are applicable to the detection of 4-CPFSK signal under the DMR digital cluster system standard; This technology belongs to the incoherent detection method; Do not need this locality that the carrier information of homophase is provided, and made full use of the Memorability of CPM signal in the judgement amount, overcome traditional poor problem of incoherent detection performance; Simultaneously; Compare with traditional many symbol detection, contain less matched filter number and Viterbi status number in the receiver of this algorithm correspondence, reduced the complexity of receiver.

Description of drawings

Fig. 1 is the schematic diagram of incoherent many symbol detection band receiver of base;

The corresponding impulse response function waveform of three matched filters in Fig. 2 incoherent detection receiver;

The different watch window N of Fig. 3, noncoherent receiver is to the detection performance of 4-CPFSK signal.

Embodiment

The incoherent detection method based on 4-CPFSK signal under the DMR standard that the present invention proposes realizes in the Matlab simulated environment, through correctness and the validity that the analysis verification of ber curve is invented.Fig. 1 has provided the schematic diagram of many symbol detection band receiver of base, and Fig. 2 has provided three impulse response function waveforms that matched filter is corresponding, and Fig. 3 has provided the detection performance of the noncoherent receiver under the different watch windows.Below in conjunction with the schematic diagram of many symbol detection band receiver of base, introduce the concrete steps that algorithm is realized:

Step 1:, define incoherent many symbol detection judgement amount of signal according to maximum-likelihood criterion.

Behind the 4-CPFSK signal process white Gaussian noise channel (AWGN), the baseband signal that receives at receiving terminal is r (t), and getting each observation length is NT, and in the time, the likelihood function of signal can be expressed as at (n-N) T≤t≤nT,

$p\left(r\left(t\right)\right|s\left(t\right),\mathrm{\&theta;})=\mathrm{Fexp}\{-\frac{1}{{\mathrm{<figure-callout\; id="0"\; label="N"\; filenames="HSA00000700521300012.png"\; state="\{\{state\}\}">N</figure-callout>}}_0}{\&Integral;}_{(n-N)T}^{\mathrm{nT}}{|r\left(t\right)-s\left(t\right)e^{\mathrm{j\&theta;}}|}^2\mathrm{dt}\}\backslash *\mathrm{MERGEFORMAT}---\left(1\right)$

Wherein, s (t) is a reference signal, and θ is the unknown phase that channel is introduced, in that (π evenly distributes in π), and F is a constant.Because the existence of unknown phase θ can't be carried out the comparison of likelihood function value, need get average about unknown phase to likelihood function, the cancellation unknown parameter is to the influence of likelihood function.Receiver structure in order to obtain to simplify is replaced reference signal s (t) with the linear breakdown of 4-CPFSK signal simultaneously, finally obtain incoherent many symbol detection judgement amount to do,

$\mathrm{\&Lambda;}=\underset{n}{\mathrm{\&Sigma;}}{\mathrm{\&lambda;}}_n={|\underset{i=0}{\overset{\mathrm{<figure-callout\; id="0"\; label="N\; c"\; filenames="HSA00000700521300012.png"\; state="\{\{state\}\}">N</figure-callout>}}{\mathrm{\&Sigma;}}}{c}_{}^{}{}_{0,i}^{\left(n\right)}{\mathrm{<figure-callout\; id="0"\; label="r"\; filenames="HSA00000700521300012.png"\; state="\{\{state\}\}">r</figure-callout>}}_{0,n-i}+\underset{k=1}{\overset{2}{\mathrm{\&Sigma;}}}\underset{i=0}{\overset{N-1}{\mathrm{\&Sigma;}}}{c}_{k,i}^{\left(n\right)}{r}_{k,n-i}|}^2\backslash *\mathrm{MERGEFORMAT}---\left(2\right)$

In the formula, r _{K, n}Be to receive signal through the sampled value behind the matched filter, N is many symbols watch window, and it is big more that its value is provided with, and the performance of detection is good more,

Be about sending signal { α _N-Nα _nFunction, its value can utilize following recurrence formula to try to achieve,

$\begin{array}{c}{\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="HSA00000700521300012.png"\; state="\{\{state\}\}">c</figure-callout>}}_{\mathrm{0,0}}^{\left(n\right)}=1;{\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="HSA00000700521300012.png"\; state="\{\{state\}\}">c</figure-callout>}}_{0,i+1}^{\left(n\right)}={\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="HSA00000700521300012.png"\; state="\{\{state\}\}">c</figure-callout>}}_{0,i}^{\left(n\right)}\exp \left(\mathrm{j\&pi;h}{\mathrm{\&alpha;}}_{n-1}\right);\\ {\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="HSA00000700521300011.png,HSA00000700521300012.png"\; state="\{\{state\}\}">c</figure-callout>}}_{1,i}^{\left(n\right)}={\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="HSA00000700521300012.png"\; state="\{\{state\}\}">c</figure-callout>}}_{0,i}^{\left(n\right)}\exp \left(\mathrm{j\&pi;h}{\mathrm{\&gamma;}}_{0,n-i}\right);{\mathrm{<figure-callout\; id="2"\; label="c"\; filenames="HSA00000700521300012.png"\; state="\{\{state\}\}">c</figure-callout>}}_{2,i}^{\left(n\right)}={\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="HSA00000700521300012.png"\; state="\{\{state\}\}">c</figure-callout>}}_{0,i}^{\left(n\right)}\exp \left(\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="HSA00000700521300012.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;h}{\mathrm{\&gamma;}}_{1,n-i}\right)\end{array}\backslash *\mathrm{MERGEFORMAT}---\left(3\right)$

Wherein, α _n=γ _{N, 0}+ 2 γ _{N, 1}λ _nBe the branch metric in the Viterbi computing, each calculating needs N+1 r _{0, n}With N r _{1, n}, r _{2, n}Participate in, clearly, branch metric λ _nBe not only to lean on the signal of a symbol time to go decision, but utilize the signal of a plurality of symbol times to improve the performance of detection.

Step 2: the design of three front end matched filters in many symbol detection receiver.

In incoherent many symbol detection receiver matched filter corresponding the linearity of 4-CPFSK signal decompose real waveform.According to the Mengali algorithm, can be with quarternary information with two binary information representations, i.e. α _n=γ _{N, 0}+ 2 γ _{N, 1}, α wherein _nBe quaternary information, γ _{N, 0}And γ _{N, 1}Be binary information, in { value in the ± 1}.According to the linear decomposition method of binary signal, obtain the amplitude modulation(PAM) pulse of the corresponding signal component of binary message, wherein again Be binary message γ _{N, 0}First waveform after the linear decomposition of corresponding signal component,

Be γ _{N, 1}First waveform after the linear decomposition of corresponding signal component.Then these waveform combination are got up to obtain the corresponding linearity of 4-CPFSK signal and decompose real waveform, through calculating first filter h in incoherent many symbol detection receiver ₀(t) do

With

Product, second filter h ₁(t) do

With

Product, the 3rd filter h ₂(t) do

With

Product.

Step 3: baseband signal through matched filter filtering, is input to the Viterbi arithmetic unit with the time point sampled value, chooses the maximum information sequence of judgement amount as the output result.

In order to carry out the calculating of judgement amount, at first need baseband receiving signals be passed through matched filter

0≤k≤2, and sample at the peak-peak point, sampled value r obtained _{K, n}

The process of incoherent many symbol detection is to choose to make the maximum information of judgement amount Λ as output, when selecting the maximum situation of judgement amount, need utilize viterbi algorithm.When watch window being set being N, can choose { α _N-Nα _N-1As state, α _nAs input, the Viterbi state transition diagram that draws contains 4 in the state diagram ^NIndividual state node, each node have four individual path inputs.Sampled value is input to the Viterbi arithmetic unit, carries out the Viterbi computing, concrete process is following:

1. branch metric calculation: at nT constantly, according to branch metric λ _nExpression formula, to each state node, calculate to get into the branch metric of four branches of node.

2. path metric upgrades: according to incoherent detection judgement amount Λ, and the renewal of carrying out path metric of the branch metric that on original arrival nT survivor path basis constantly, 1. obtains with step.At this moment, each state node to moment nT will obtain four path metric values.

3. confirm survivor path:, and keep maximum institute respective path, other three paths of cancellation simultaneously to the path metric value after each its four renewal of state node comparison of moment nT.After all state nodes are all accomplished aforesaid operations, each node paths of will only surviving.Keep survivor path data and path metric data.

4. obtain to detect data: symbol time of every mistake, repeat above-mentioned steps and 1. arrive 3., up to EOS.Select the path with maximum path metric then, the maximum likelihood path of ascending the throne is recalled the data that preserve in the path, obtains detecting data.

The emulation of Fig. 3 different watch window N, noncoherent receiver can find out from the ber curve of receiver that to the detection performance of 4-CPFSK signal along with the increase of watch window N, the detection performance of receiver improves gradually.

The content of not doing in the application of the present invention to describe in detail belongs to this area professional and technical personnel's known prior art.

Claims (2)

1. an incoherent detection that is applicable to the DMR digital clustering communication system is technological, and the step that is adopted is:

Step 1: according to maximum-likelihood criterion; Define incoherent many symbol detection judgement amount of signal; Under additive white Gaussian noise channel, can calculate the likelihood function that a plurality of symbol times receive signal, the method for taking to make even equal to the phase place in the likelihood function is eliminated the influence of unknown parameter to judgement; Utilize the primary signal of the linear breakdown alternative reference of 4-CPFSK signal simultaneously, the back of deriving obtains final incoherent many symbol detection judgement amount;

Step 2: the design of three front end matched filters in many symbol detection receiver; At first with quarternary information { ± 1; ± 3} is expressed as two binary message { ± 1}; Accomplish the linear amplitude modulation(PAM) impulse waveform of decomposing of the corresponding signal component of two binary messages, with the real waveform of these decomposition according to principle combinations together, the linearity that finally obtains 4-CPFSK is decomposed real waveform then;

Step 3: baseband receiving signals through filter match filtering, is input to the Viterbi arithmetic unit with the time point sampled value, chooses the maximum information sequence of judgement amount as the output result.

2. according to a kind of incoherent detection technology that is applicable to the DMR digital clustering communication system described in the claim 1, it is characterized in that:

(1) incoherent many symbol detection algorithm of the present invention's employing is once observed the detection performance of the reception signal raising system of a plurality of symbol times;

(2) according to the Mengali algorithm; Three amplitude modulation(PAM) pulses can have been represented the substandard 4-CPFSK signal of DMR digital cluster communication; The present invention utilizes the linear signal of decomposition to substitute original reference signal simplified receiver structure, and receiver is made up of three matched filters;

(3) when the situation of carrying out the judgement amount maximum is chosen, along with the increase that detects sequence, detector need be selected a kind of situation of judgement amount maximum from very many possibilities, implement more complicated, can select { α _N-Nα _N-1As state, α _nAs input, the Viterbi state transition diagram that draws selects for use viterbi algorithm to simplify the process that detects.

Images