Virtual Testing Area for Solving EMC Problems of Spatially Distributed Radiosystems based on Automated Double-Frequency Test System

Transcription

1 Virtual Testg Area for Solvg EMC roblems of Spatially Distributed Radiosystems based on Automated Double-Frequency Test System Vladimir I. Mordachev, Eugene V. Sevich Electromagnetic Compatibility R&D Laboratory Belarusian State niversity of Informatics and Radioelectronics (BSIR) 6,.Brovi st., Ms 0013, Belarus Abstract A new technology named as Virtual Testg Area and tended for electromagnetic compatibility (EMC) analysis complexes of radio systems is presented. This technology is based on semi-physical EMC modelg with an expedient ratio of mathematical and physical modelg of (a) electromagnetic environment (EME) and (b) behavior of radio receivers complicated EME. The proposed technology has the followg advantages: (a) high formativity of the Automated doublefrequency testg technique (ADFTT), which is used for characterization of receivers, (b) high efficiency of the Discrete nonlear analysis (DNA) technique, which is used for behavior simulation of receivers, (c) high objectivity of EME modelg with the use of digital area maps and geoformation systems (GIS), (d) efficiency and low cost of software-controlled physical modelg of receivers laboratory conditions by usg the equipment of Automated Double-Frequency Test System (ADFTS). Automated Double-Frequency Test System; Discrete Nonlear Analysis; hysical Modelg of Radio Receivers; System-Level EMC Analysis I. INTRODCTION ADFTT is one of the most effective technologies for radio receiver examation durg the developg and solvg of EMC problems of the radiosystems worg the most severe EME presented onboard or ground complexes of radiosystems [1], [], [3]. Systems usg this technique givg real opportunities of automated detection, identification and measurement of parameters for the ma channel and all spurious and termediate radio receiver paths and responses, through which terference can fluence any radio devices, and of radio receiver susceptibility to nonlear effects: blocg, cross modulation, all types and orders of termodulation, etc. ADFTT allow to collect the full formation on selective and nonlear properties of radio receivers, particular, to measure its put nonlearity up to parameters of nonlearity of rather high (15...5) orders. This maes possible essentially to crease quality of the EMC analysis severe EME usg discrete EMC-analysis and prediction technique [4], [5], [6], [7], [8]. Functionalities of variants [1], [] of ADFTT realization were substantially limited by restricted functional and metrological features of used radiomeasurg devices of previous generations, especially on frequencies above 1- GHz, and also by restricted features of the software for the EMC analysis based on usg tests results. Modern ADFTT features under the solvg of EMC problems of radiosystems can be considerably expanded due to use of Modern tellectual program-controlled vector measurg apparatus - RF signal generators havg a mode of ternal baseband generation on multicarrier mode with different types and parameters of digital modulation (SK QAM FSK, MSK) and spectrum analyzers havg a mode of signal distortion measurement, Modern technologies of nonlear discrete behavior simulation of radioreceivers [4], [5], [6], [7], [8] usg results of its ADFTT testg, Specialized GIS and expert systems for EMC modelg spatially-distributed and local on-board or onground assemblage of radiosystems [7], [9]. Consequently, the advanced ADFTS becomes the effective Virtual Testg Area providg a feature of physical modelg of fluence on the RT of the totality of spatially distributed and/or locally aggregated radiosystems, i.e., a feature of practical checg their mutual EMC conditions. The paper is organized as follows. In Section II, we troduce the advanced ADFTS structure and peculiarities. A technique for extractg the receiver-under-test (RT) model from ADFTS measurements is proposed Section III. The Discrete nonlear analysis (DNA) technique, which is used for nonlear behavior simulation of RT, is described

2 Section IV. A technique of RT modelg and EMC analysis by usg ADFTS as a Virtual Testg Area is presented Section V. II. ADFTS STRCTRE AND ECLIARITIES The basic structure of the ADFTS advanced version used for Virtual Testg Area realization is shown Fig.1. Key elements of the ADFTS are the RF signal generators and ADC (analogue-digital converter). Computer is used primary for soft control by test signals and to evaluate signal parameters at the put of the RT. The technology of radiolocation testg of receivers with the use of ADFTS cludes detection, identification and measurg of parameters and characteristics of all paths and phenomena (which can affect receiver operation under the conditions of specified (predicted) maximum signal levels and over all possible put signal frequencies) cludg spurious response paths, all paths (types) of two-signal termodulation, blocg, crosstal, excitation of put stages under the fluence of strong -of-band signals, locg of the local oscillator frequency by an put signal, etc. On the first stage, the form of the double frequency amplitude characteristic of the receiver-under-test (cludg and cross-sections of this characteristics) is analyzed. This characteristic H ( f 1, f ) has a dependence H = const, (1) 1 ( f ) = 1 f f1, f = const where is the put signal level when two test signals at frequencies f 1 and f with levels 1, correspondgly are applied to the receiver put; as a rule, =. 1 = max This stage is completed by recordg and displayg coordates ( f 1, f ) one or several followg cross sections of the double-frequency amplitude characteristic H ( f 1, f ) : W ( f f ) sgn H ( f f ) { } i 1 ti = 1, ti, () at the specified threshold levels ti, i = 1,,..., where the label sgn means signum function. These levels are selected so that they exceed the level of the ternal noise of the receiver at its put accordance with the accepted criteria used for determation of the receiver ma channel sensitivity and are also selected to test the receiver susceptibility due to spurious response paths and nonlear effects. Recorded images of cross sections of the double-frequency amplitude characteristic are, effect, fragments of the nown doublefrequency diagram (DFD) of the receiver; typical examples of such diagrams are presented []. Fig. 1. Structure of the advanced ADFTS. The ma ADFTT & ADFTS peculiarity is followg. Twodimensional (D) image of DFD is generated ( f 1, f ) D coordate system with the use of horizontal (fast) frequency scanng of the RF Generator 1 on area [ f 1m ; f1max ] with period T and frame (slow) frequency scanng of the RF Generator on area [ f m ; fmax ] with period T C = n T (as on TV raster). ractically it is possible to form a good quality DFT image usg n = les (if frequency scanng is made not very wide frequency ranges; as a rule, f ; f ] = [ f ; f ] = [ f ; ] ). [ 1m 1max m max m fmax On the fal stages, the identification and measurg of parameters and characteristics of all detected paths and phenomena RT antenna put is made. As a result all necessary and comprehensive formation concerned RT susceptibility to lear and nonlear terferences on antenna put can be extracted and used for the detailed EMC analysis (analysis of RT behavior EME limited by, [ f ; f ]} ). { max m max

3 The description of assignment of other ADFTS elements is given below. ADC is used when a RT put signal is taen from its termediate frequency (IF) put: ADC the amplification, detection (of root-mean-square type or another type, if necessarily) and analogue-to-code transformation of an IF signal are made. Frequency counters are used for simultaneous measurement of frequencies of RT IF signal and RT heterodyne signals at identification of the founded RT spurious or termodulation paths. The RF generator 3 is required for desired signal jection on RT antenna put, if necessary. Additional RF generators are used, if necessary, for auxiliary signals jection on Object nder Test (a reference signal, a signal of local oscillator at mixers testg, etc.), and also for modelg of terference signals alongside with RF generators 1, at physical EMC modelg (RT behavior modelg expected EME). The spectrum analyzer and oscilloscope are used at the detailed analysis of the RT put signal, cludg measurement of a distortion of a desired signal under terference fluence. The Summator Σ can be constructed usg typical coaxial summators, but realization of summation of test signals by field with use of measurg antennas and RF amplifiers is also possible (for example, at tests of receivg active phased-array antennas anechoic chambers or on the open area). The important prelimary procedure of ADFTT - metrological calibration of networs for submission of test signals from puts of RF generators on RT antenna put (between the puts and the put of Summator Σ ) a wide frequency range Df = [ f m ; fmax ] which is carried before the begng of tests. This calibration is made with use of RF generators and spectrum analyzer. Durg performance of the calibration procedure measurement of amplitude-frequency characteristics (AFC) A j ( f ), j = 1,... n of each of n ways of test signals passage between each of n puts of the Summator Σ and its put all frequency range of tests and the EMC analysis is made. Application of ADFTS measurg modules of the top level (RF generators, spectrum analyzer) providg a very high accuracy of stallation and measurement of levels of signals and tag to account AFC A j ( f ) of the each put of Summator Σ the total error of stallation of levels of RF signals on RT put can be less than 1 db. It provides the high accuracy of the control of RT EMC characteristics, and high objectivity and accuracy of mathematical and physical RT behavior modelg & simulation given EME (EMC modelg) at fal stages of tests and the EMC analysis of RT with use of advanced ADFTS on Fig.1 as a Virtual Testg Area. III. RT IDENTIFICATION SING ADFTS The Filter Nonlearity Filter model (FNF) (which is also referred to as Typical radio-engeerg stage [10], Wiener Hammerste model [11], Nonlear sensor [8]) is widely applied for behavioral simulation of nonlear effects receivg front-ends. This model (ref. Fig. ) taes to account the followg properties [4]: frequency selectivity of the receiver s put circuit (the 1st lear filter), nonlearity of the front-end (the memoryless nonlearity), and termediate frequency (IF) selectivity (the nd lear filter). The quality of a behavioral model is determed not only by the structure of the model but also by the identification (i.e., extraction of model parameters from measurements) technique. The FNF model identification methods developed for application control systems (see references given [10], [11]) not always yield good results when applied to simulation of radio equipment [1]. Therefore, a technique for identification of the FNF model for a radio receiver is proposed below; this technique is tended for EMC analysis, and it has the followg important features. 1) The ability of model identification a wide frequency band (usually, the decade up and the decade down from the tung frequency of the receiver) and a wide dynamic range (not narrower than the receiver s dynamic range, which can achieve 10 db). ) The applicability even to a receiver that do not provide access to its ternal structure (e.g., the tegrated-circuit receiver) and do not have the IF put. 3) The technique is based on standardized characteristics of the receiver and on standardized measurement procedures. The proposed technique of identification considers the FNF model as a generalization of the traditional lear model (the receiver susceptibility characteristic the frequency doma [13, Fig. 4.], [14]). The technique consists of three stages: 1) synthesis of the lear model for the receiver, ) identification of the put filter, and 3) identification of the nonlearity. Note that one can synthesize more detailed model than the FNF model if he has access to the ternal structure of the receiver. This can be done as identification of models for the units (amplifiers, mixers, etc.) of the receiver [4], [5], [6], [7]. In Lear filter 1 H1 ( f ) Memoryless nonlearity u ( u ) Lear filter ( f ) H Fig.. Behavioral model of radio receiver. Out

4 A. Synthesis of lear model The lear model is extracted from the frequency-doma susceptibility characteristic (FDSC) η ( f ) of the receiver. This characteristic may be measured by the two-signal technique or one-signal technique (the latter can be used if the former is not appropriate) [15, MIL-STD-461/46/461F method CS04/CS104]. As shown [14], the calculation of the tegrated terference marg IM (which is the criterion of EMC) from the FDSC η ( f ) by the formula = IM [ ( f )/ η ( f )] (3) (where ( f ) is the power of -th undesired signal at the receiver put; f denotes its carrier frequency) is equivalent to the use of the lear filter as the receiver model and to the comparison of the total power, Σ of terference at the filter put with the threshold ~ η : Σ IM = ~, = ~ 1 ( f ), (4) η η ( f ) = H ( f ) ( f ), (5) where H ( f ) is the power transfer function of the filter. As follows from (4) and (5), the values of H ( f ) and ~ η are accurate with a constant multiplier K : multiplyg both H ( f ) and ~ η by K leaves IM (4) unaltered. It is convenient to defe H ( f ) and ~ η so that the transfer function H ( f ) is normalized to unity at the tung frequency f 0 of the receiver, i.e., H ( f 0 ) 1 : H ( f ) = ~ η / η ( f ) ; ~ η = ( ). (6) η The synthesized lear model of the receiver (4), (5), (6) describes not only lear effects (selectivity) the IF circuit but also a number of nonlear effects the preselector and frequency converter(s): spurious responses, reciprocal mixg, desensitization (the last two effects are accounted by the model if the FDSC η ( f ) was measured by the two-signal technique). The ability to account for modulated terferg signals by performg the FDSC measurements with the use of such signals [13], [14], [15, MIL-STD-449D method CS114] is the second advantage of the model. Let us consider the drawbac of the lear model: the model uses the prciple of terference superposition (4) and, therefore, it does not account for termodulation. As a rule, the lear model describes the receiver adequately the followg widespread situation: one powerful unwanted signal domates the EME; but even this situation an termodulation (between the domatg and a wea unwanted signals) may arise and suppress the desired signal. To correct the mentioned drawbac, the nonlearity should be troduced to the structure of the model (ref. Fig. ), and f 0 the selectivity of the lear model s filter should be divided between the put and put filters of the FNF model. When the FNF model of the receiver is simulated by the DNA technique (ref. Section IV), the nonlearity is modeled the time doma. Therefore, it is necessary to represent the lear model (5) terms of the put and put voltages: ( f ) = R ( f ) ( f ), (7) ( f ) = H ( f ) ( f ), (8) ( f ) = ( f ) / [ R ( f )], (9) where H ( f ) is the voltage transfer function of the filter, i.e., the frequency response (FR), calculated as H ( f ) = H ( f ) R ( f )/ R ( f ) ; (10) R ( f ) and R ( f ) are the put and put resistances of the receiver, respectively. In prciple, these resistances may be arbitrary because their values have no effect on the EMC criterion calculated by substitution of (7), (8), (9), and (10) (4). Nevertheless it is desirable to use realistic values of R ( f ) and R ( f ) if such values are nown. B. Identification of put filter The frequency response of the put filter is extracted from the third-order-termodulation selectivity characteristic (IM3SC) of the receiver [15, MIL-STD-449D method CS110], [13, p. 4.49], [16, GOST , GOST ]. The choice of IM3SC (and not the desensitization selectivity characteristic) is based on the worst-case approach to EMC analysis: termodulation and desensitization may arise different stages of the receiver, but usually termodulation emerges from a lower power of undesired signals and is therefore more dangerous. The IM3SC may be measured by the two-signal (or threesignal, if necessary) technique [15, MIL-STD-449D method CS110]. Note that the full IM3SC is not measured the most of standard tests: just the fact of the IM3SC location above the specified limitg characteristic is checed one or several pots [15, MIL-STD-461/46/461F method CS03/CS103], [16, GOST , GOST 15-86]. Let us consider the IM3SC of the FNF model. Sce the level of terference at the receiver put is mataed low durg IM3SC measurements (equivalent desired signal at the tung frequency is equal or near to the receiver sensitivity), the contribution of nonlear terms of order higher than 3 to IM3SC can be neglected and the stantaneous transfer characteristic (ITC) of the memoryless nonlearity (MNL) (ref. Fig. ) can be represented as u 3 ( u ) a1 u + a u + a3 u =, (11)

5 where u and u are the stantaneous voltages at the put and put of the nonlearity, respectively; a 1, a, a3 are constant coefficients. Feedg the two-tone signal to the FNF model put and tag to account the FR H1 ( f ) of the put filter, ITC (11) of the MNL, and the FR H ( f ) of the put filter, we fd the voltage amplitude ( f 0 ) of the termodulation terference (at the tung frequency f 0 of the receiver) at the put of the FNF model ( f0 0 nl f0 ) = H ( f ) ( ), (1) 3a3 ) = [ H1 ( f N ) ( f N)] H ( f F ) ( f ), (13) 4 nl ( f0 1 F f N = f 0 + f ; f F = f0 + f, (14) where nl ( f 0 ) represents the termodulation voltage amplitude at the MNL put; f N is the frequency of the put signal tone nearest to f 0 ; f F is the frequency of the put tone farthest from f 0 ; ( f N ) and ( f F ) stand for the voltage amplitudes of the put tones, they are calculated by (7); f is the difference between f N and f 0. The terference power at the FNF model put is calculated by substitution of (1) to (9). By the defition [15, MIL-STD-449D method CS110], [13, p. 4.46], [16, GOST , GOST ], the IM3SC is a dependence ( f N ) of the put tone power at the frequency nearest to f 0 on the frequency f N of this tone under the followg conditions: the termodulation terference power at the receiver put is mataed constant and the put tones are equal power. Consequently, as it follows from (1) and (9), the terference voltage amplitude (13) at the MNL put is constant durg the computation of the IM3SC for the FNF model. This maes it possible to compute the FR H1 ( f ) of the FNF model s put filter on the base of (13) and (7) by the followg algorithm. 1) Sce the put selectivity of the receiver is many times less than the selectivity of its IF circuit, assume that for mimal difference f 1 the equality H f ) H ( f ) 1 (15) 1 ( N = 1 1 F1 = holds and calculate a parameter β from (13): β ( f ) / (3a ) = [ ( f )] ( f ), (16) 4 nl 0 3 N1 F1 where the values of ( ) are computed by (7). ) At each further (i-th) step, it is necessary to crease the frequency difference two times as compared with the previous step: i i 1 fi 1 = f1, i = f =,3,... M. (17) R Then, as it follows from (17) and (14), f, (18) N = f0 + fi = f0 + f i 1 = f i F i 1 which maes it possible to express the FR of the put filter at the frequency f from (13) as H F i β / {[ H ( f ) ( f )] ( f )}, (19) 1 ( f F i) = 1 F i 1 F i 1 F i where the values of ( ) are computed by (7). 3) For the negative differences f (14), to calculate the left branch of the FR H1 ( f ) : sce the parameter β is already nown, it is possible to compute the recursion (17), (19) under the conditions H = 1,,... M. (0) 1 ( f N ) H1 ( f F 0) = 1; i = 1 4) To combe and terpolate the calculated values of the left and right branches of the FR; as a result, we obta the contuous function H ( ). 1 f 5) Outside the frequency terval of the IM3SC measurement, it is necessary, accordg to the worst-case approach, to tae the FR of the put filter equal to its value at the bound of the mentioned terval: H1 ( fm), f < f m = f0 + f M ; L H 1 ( f ) = (1) H1 ( fmax), f > fmax = f0 + f M. R After the identification of the put filter, one can compute the FR of the FNF model s put filter as H ( f ) = H ( f ) / H1 ( f ), () where H ( f ) is the FR (10) of the lear model. C. Identification of nonlearity As a result of the nonlearity identification, the high-order polynomial model of the ITF is synthesized; such model ensure adequate description of both the baced-off region with a mild nonlearity, dangerous terms of termodulation emergence, and the saturation region havg essential nonlearity, hazardous terms of desensitization. If the RT does not have the IF put, the model of the nonlearity should be extracted from measurements of the spurious-free dynamic ranges for termodulation of different orders and for desensitization [17], [18]; but if the IF put is available, then more exact model for the odd part of the ITF can be extracted from measurements of the sgle-tone amplitude-to-amplitude characteristic ( adjacent-signal region) and the two-tone amplitude-toamplitude characteristics for termodulation of odd orders [18], [5], [19]. IV. NONLINEAR BEHAVIOR SIMLATION OF RT For behavioral simulation of radio receivers operatg severe electromagnetic environment (EME), the discrete nonlear analysis (DNA) technology has been developed and L

6 successfully applied [4], [5], [6]. This technology has a number of essential features to analyze electromagnetic compatibility: 1) simulation wide frequency and dynamic ranges, ) account for combed fluence of all the fundamental types of nonlear effects (termodulation, desensitization, crossmodulation, spurious responses, reciprocal mixg), 3) high computational efficiency (the DNA s computational advantage over traditional techniques grows rapidly with creasg EME complexity and with creasg order of the nonlear effects). Commercial implementation of the DNA the software EMC-Analyzer [7] has also the followg important functions (see references given [6]): 4) automatic search of nonlear terference sources by the dichotomous method, 5) estimation of the signal-to-terference ratio at any pot of the receiver s nonlear model. The DNA technology is based on the followg prciples: 1) utilization of discrete models for EME and for composite signals the receiver s front-end; ) representation of the receiver model as lear filters and memoryless nonlearities connected series (or parallel, or both); 3) application of the efficient modelg methods for the lear and nonlear transformations (lear filters are modeled the frequency doma, memoryless nonlearities are modeled the time doma, the transition from the time doma to the frequency doma and vice versa is made with the use of the fast Fourier transforms); 4) use of polynomial models for the memoryless nonlearities (these models ensure a controllable crease of a signal s bandwidth as a result of the nonlear transformation and mae it possible to achieve a high dynamic range of the nonlear analysis up to 300 db). An overview of the current state of the art the DNA technology is presented [6]. V. RT EMC ANALYSIS SING ADFTS AS A VIRTAL TESTING AREA Virtual Testg Area allows effectively and with high accuracy to solve the followg EMC problems at a system level: 1) EMC analysis of a large quantity of probable scenarios of RT allocation on area with given radio electronic environment (REE). ) The analysis of probable RT affection by terferences at various scenarios of radiosystems area allocations (scenarios of REE formation, for example, variants of area allocations of base and radio-relay stations of a cellular networ). The itial stage of the given wors is formation of EME mathematical model on RT put as ensemble of signals with the specified parameters (levels, carrier frequencies, types and parameters of modulation). This EME model can be received several ways: a) By experimental EME measurement places of probable RT allocation. b) By calculations usg formation on characteristics of electromagnetic radiation and coordates of area allocation of radio transmitters - probable sources of terferences, and also IT-R radio waves propagation (RW) models [9] and digital area maps. In the latter case at EMC analysis for various scenarios of radiosystems area allocation the EME model on RT put for each scenario is formed. Generation of each EME model is carried with use of geoformation technologies and clude the followg procedures (Fig. 3): 1) Modelg of variant (scenario) of area allocation of transmitters of radiosystems potential terference sources. ) Calculation of levels of electromagnetic fields of these radio transmitters a pot of RT antenna area allocation with use of RW models [9] tag to account surface relief, vegetation, precipitations and other factors. 3) Calculation of levels of radio signals from environmental radiosystems on RT put with use of the data on spatial, frequency and polarizg selectivity of RT antenna, formg of EME model as ensemble of put signals with specified parameters. 4) Detection of group of the most dangerous signals, capable to be the cause of terference for RT performance. After that, the EMC problems mentioned above can be analyzed and solved at various levels of virtuality of reproduction of radiosystems terference teractions: a) Mathematically with use of technology of RT discrete nonlear behavior simulation given EME (ref. Section IV). b) hysically by carryg- the physical EME model generated by RF generators 1,... n on RT put (on an put of the Summator Σ Fig. 1). Fig. 3. Generation of electromagnetic environment model.

7 On this stage all numerous ensemble of signals on RT put with losg of objectivity of the EMC analysis can be replaced with the limited number of the most dangerous terferences from this ensemble of signals. This the most dangerous terferences may be formed on RT put by RF generators, cluded ADFTS. As a rule, even if necessary to generate for RT a desired and/or a reference signals for qualitative RT physical behavior modelg and the objective EMC analysis given EME it is necessary to have ADFTS structure no more than 4-5 RF generators ( n 5), because practice termodulation is formed by no more than two - three most powerful put signals. Such tests many cases allow to replace real radiosystems ground tests on EMC at which real area allocation and operation of large complex of real systems is made, with much less expensive laboratory tests with use of Virtual Testg Area presented above. In this cases detailedness and practical utility of EMC estimations appears better due to checg of all probable scenarios and variants of radiosystems application of their regional or local (onboard, ground) aggregation. VI. CONCLSION Efficiency of the Virtual Testg Area technology is confirmed by results of its practical use and defed by advantages of its basic elements: origal technology of testg and identification of EMC characteristics of radio receivers (receptors of terference) based on ADFTS, origal technology and software for the discrete nonlear EMC analysis and behavior simulation of receivers operatg severe EME, the technique of EME modelg with use of digital area maps and specialized GIS, and fal physical modelg of receivers presence of a set of the most dangerous terferers laboratory conditions by usg the ADFTS equipment. Materials presented this paper generalize the authors experience the field of ADFTS [1], [], EMC-Analyzer expert system [5], [6], [7], [8], and specialized GIS for EMC development and practical usage for solvg system-level EMC problems. REFERENCES [1] Aporovich A.F., Mordachev V.I. Functional possibilities of the EMC characteristics monitorg of electronic apparatus by the two-frequency probg method // IX Int. Wroclaw Symp. EMC, 1988, ( Russian). [] Mordachev V.I. Automated double-frequency testg technique for mappg receiver terference responses // IEEE Trans. on EMC, Vol.4, No, May 000, [3] orter J.D., Billo R.E., Micle M.H. Effect of active terference on the performance of radio frequency identification systems // Int. J. Radio Frequency Identification Technology and Applications, Vol. 1, No. 1, 006, [4] Mordachev V.I. Express analysis of electromagnetic compatibility of radio electronic equipment with the use of discrete models of terference and fast Fourier transform // IX Int. Wroclaw Symp. EMC, 1988, ( Russian) [5] Loya S.L., Mosig J.R. New behavioral-level simulation technique for RF / Microwave applications. art I: Basic concepts // Intern. J. RF and Microwave CAE Vol. 10, No [6] Mordachev V.I., Sevich E.V. EMC-Analyzer expert system: improvement of IEMCA models // XIX Int. Wroclaw Symp. EMC [7] EMC-Analyzer. Mathematical models and algorithms of electromagnetic compatibility analysis and prediction software complex. Ms, 009. [8] V.I.Mordachev,.A.Litvo. Nonlear sensor method for EMC/EMI analysis severe electromagnetic environment usg EMC-Analyzer expert system // XVII Int. Wroclaw Symp. EMC, 004, [9] IT-R Recommendations, Series (Rec. IT-R and others), Geneva, IT, 008. [10] Sverunov Y.D. Identification and quality control of radio-electronic systems nonlear elements ( Russian). Moscow, [11] Crama., Schouens J. Computg an itial estimate of a Wiener Hammerste system with a random phase multise excitation // IEEE Trans. Instrumentation and Measurement Vol. 54, No [1] Crama., Rola Y. Broad-band measurement and identification of a Wiener Hammerste model for an RF amplifier // 60th Automatic RF Techniques Group Conf. Washgton, DC, [13] A handboo series on electromagnetic terference and compatibility, Vol.7: Duff W.G. Electromagnetic compatibility telecommunications. Interference Control Technologies, Inc., Gaesville, VA, [14] earlman R.A. hysical terpretation of the IEMCA tegrated EMI marg // 1978 IEEE Symp. EMC [15] Standards (MIL-STDs), SA Department of Defense, Washgton, 010. [16] State Standards (GOSTs), (former) SSR State Committee on Standards, Moscow, [17] Cheremisov I.D., Loya S.L., Mordachev V.I. Synthesis of the polynomial model of nonlear elements based on termodulation dynamic ranges // 3-rd Int. Conf. on Telecomm. modern satellite, cable and broadcastg services. Nis, Yugoslavia, 1997, [18] Loya S.L., Cheremisov I.D. Validation of the high-order polynomial models used behavioral-level simulation // 4-th Int. Conf. on Telecomm. modern satellite, cable and broadcast. services. Nis, Yugoslavia, 1999, [19] Sevich Е.V. AM-M conversion simulation technique for discrete nonlear analysis of electromagnetic compatibility // 8-th Intern. Symp. on EMC and electromagnetic ecology. St. etersburg, June 16 19, 009, Vladimir I. Mordachev (M 96) was born Vitebs, Belarus. He received the h.d. degree (1984) Radio Engeerg from the Ms Radio Engeerg Institute, Ms, Belarus. His research terests clude spectrum management, wireless communications and networs, electromagnetic compatibility and terference, wireless networ planng, computer-aided analysis and design, cellular networs system ecology, RF systems modelg and simulation. He is extensively volved consultg to wireless networ operators, dustry and the local government. V. Mordachev is a Head of the Electromagnetic Compatibility R&D Laboratory at the Belarusian State niversity of Informatics and Radioelectronics. Eugene V. Sevich was born Ms, Belarus, He is currently engaged as Chief designer of EMC-Analyzer and DFTS projects EMC R&D Laboratory of Belarusian State niversity of Informatics and Radioelectronics. Ma areas of terest: electromagnetic compatibility and terference, computer-aided analysis and design, wireless communications, and signal processg.