Based On N Prominent Spectral Peak Index Of Typical Research And Implementation Of Digital Modulation Signal Blind Identification

Over the past years, modulation recognition technique has been an importantresearch topic in the field of wireless communication. Whether in the militaryinvestigation or civil communication, there are many application of modulationidentification technique. At present, methods of modulation identification can mainly bedivided into two categories: modulation recognition based on likelihood ratio andmodulation recognition based on pattern recognition. In the first method, test statisticsare complex, and it also needs related prior knowledge of signal; but the second methodis relative simple, which usually use FFT, or cycle spectrum, or higher order cumulativeto extract signal features, and then complete recognition with decision tree, or neuralnetwork, or support vector machine. However, in actual environment, it is almostimpossible to get prior knowledge of the received signal. Blind modulation recognitionproblems will be more difficult.This article carried out a study and realization on blind identification of typicaldigital modulation signals, based on N peak convex index, including2ASK2FSK,4FSK,8FSK, BPSK, MSK six kinds of signals. The full text includes three main parts:algorithm design and simulation, software radio implementation, performance testing.Aritcle is arranged as follows:Chapter one and chapter two introduce the present situation and development ofmodulation recognition technology.Third chapter is the core part of this article. It figured out the principle of how toperform the blind recognition of typical digital modulation signals with N peak convexindex, explained design process and performance simulation of this algorithm, anddiscuss performance effect caused by changes on recognition parameters, such asjudgment data points, code number, sampling rate, and spectrum overlay number. Inwhite Gaussian noise channels, when the symbol rate is0.5M Baud and code number is1638, simulations show that2ASK,2FSK in-16dB,4FSK in-11dB,8FSK in-8dB,BPSK in-1dB, MSK SNR more than3dB, recognition probability is above90%. Nooverlay system performance is better than overlay systems; the more code number, the better performance.The fourth chapter describes implementation of the blind recognition algorithmbased on N peak convex index for typical digital modulation with software radio. Thisarticle uses SDR DP (Soft Defined Radio Development Platform, SDR DP), producedby Lyrtech company in Canada. Based on their VPSS data transfer interface (VideoProcessing Subsystem) characteristics, this article designs two kinds of modulationrecognition system: the hardware overlay system and no hardware overlay system.Underlying output no hardware overlay system is the collection of64K point in timedomain data, other operations carried out by top; while in hardware overlay system, theoutput is multiple added spectrum data. Deficiency of two systems: no hardware overlaysystem has a good recognition performance, but requires for a long time; in hardwareoverlay system, identification time is short, but less recognition performance. Thechapter also carefully record the link design process, arbitrary frame add subsystem, bitmerge subsystem, upper-class GUI interface design and so on.The fifth chapter is to identify system performance testing and analysis. Testingperformance of modulation recognition system, and simulation results were compared.Hardware no overlay system and simulation (no overlap) compared: achieve theidentification probabilities above90%,2~3dB performance loss. Under the conditionsof the same parameters, hardware overlay system divide data into each section of4,096,and then add spectrum data of each section, compared to non-overlay system,1~2dBwas loosed. Non-integer multiple of the sampling lose1~2dB than the integer multipleof the sampling. This may be due to a variety of factors: noise distribution in the actualplatform, test number of test statistics, as well as the link itself point quantization error,and so on.The full text carried out algorithm design, realization, and test, combined,compared to each other, and eventually proved the feasibility of algorithms. A highlightof this article is to use N peak convex index to make blind modulation recognition anddo a hardware implementation and practical tests. This has practical value, can providereference for future engineering studies.