Available online at www.sciencedirect.com ScienceDirect Procedia Computer Science 103 (2017) 439 – 446 XIIth International Symposium «Intelligent Systems», INTELS’16, 5-7 October 2016, Moscow, Russia Features of control algorithms for frequency distribution in conditions of mass use Internet of Things V Loginov, S Pavlov*, Y Fedosenko Volga State University of Water Transport, 5, Nesterova Str Nizhny Novgorod, 603605, Russia Abstract In this article is considered the problem of operational frequency management in terms of its dynamic use, taking into account non-linear distortion in high cascades transceivers Main results contain the new method for operational management establishing, taking into account non-linear distortions in communication channels This method satisfies the modern requirements for its use in the systems of dynamic allocation of frequency resources The method improves the use of communication channels efficiency © 2017 The TheAuthors Authors.Published Publishedbyby Elsevier B.V © 2017 Elsevier B.V This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the XIIth International Symposium «Intelligent Systems» Peer-review under responsibility of the scientific committee of the XIIth International Symposium “Intelligent Systems” Keywords: frequency resource; the DSA; channels of communication; cognitive radio; non-linear distortions; Raman frequencies Introduction The information society formation is associated with the development of Internet technologies Since 2010, the Internet of Things technology (IoT) has been rapidly developing1 It is based on the concept of physical computer network facilities equipped with built-in means to provide their interaction with each other and (or) with the external environment This concept considers the organization of such networks as a phenomenon that can rearrange social and economic processes Wireless telecommunications such as a widespread use of radio frequency identification obtain high-level importance for IoT implementation The problem of communication channels reliability is exacerbated due to the widespread transformation of the ordinary objects that surround people all over the world as well as industrial * Corresponding author E-mail address: pavlovstanislav@mail.ru 1877-0509 © 2017 The Authors Published by Elsevier B.V This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the XIIth International Symposium “Intelligent Systems” doi:10.1016/j.procs.2017.01.013 440 V Loginov et al / Procedia Computer Science 103 (2017) 439 – 446 equipment and means of materials transportation into Internet hubs The reason for this is the emergence of deficit of frequency resources available for use, the deterioration of interference environment and electromagnetic compatibility of radio telecommunication facilities The current forecasts of IoT development tendency show that the number of physical objects equipped with the means of interaction with each other and/or with the environment is expected to reach the values of approximately 1010 in the near future According to the points described above, we can assume that ensuring the functioning of IoT with the specified quality parameters has a primary importance in the process of infrastructure and wireless telecommunication development Frequency resources deficit arises due to the established method of their Authorizing Spectrum Access (ASA) The method implies the users’ provision with a fixed frequency range communication channel and the parallel use of a traditional technology, which does not allow setting operating frequency, power output and the type of modulation selection However, in large cities with a fully distributed authorized frequency resources, the average daily-unused frequencies is as high as 70% To find a possible way out of this situation, it is proposed to replace the ASA with the Dynamic Spectrum Access (DSA3), forming the core technology of the cognitive radio system (CRS2) The problem of electromagnetic compatibility provision within CRS is solved partially Its main advantage is the ability to optimize the communication channel bandwidth using signal/noise ratio at the input of the receiver Therefore, nonlinear frequency conversion interference as well as intermodulation interference on the ingress path of the radio receiver device (RRD) are not taken into account in this case This may lead to the formation of communication channel suboptimal quantities and parameters, as a consequence, that generate additional time required for their iterative adjustment procedure So, it is necessary to carry out the forecast for the signal/noise ratio at the demodulator input Fig Usual radio channel There is an approach that there is no such a procedure in prior CRS engineering implementations, moreover the time of automatic continuous retuning connection from one channel to another amounts approximately to one microsecond To forecast and evaluate signal/noise ratio in “real time” Our proposal is to set a new methodology and optimization concept that enables to build the two-signal macro model using Farey series, instead of calculating the full range of non-linear frequency conversion under random input action4 The evaluation of spurs in frequency mixer at two-signal exposure is reduced to the calc of the spectrum in the “near” area5 At the final calc stage, we find out optimal values of communication frequencies parameters, channel bandwidths and the levels of spurs at the output of nonlinear system V Loginov et al / Procedia Computer Science 103 (2017) 439 – 446 Fig Structure of CRS radio channel We conclude that this approach allows us to implement a fast algorithm to calculate the predicted values of the signal/noise ratio and the optimal values of the communication channel parameters5 This is to ensure reliability of wireless telecommunication functioning in framework of IoT technologies wider spread Stage Searching for the double Diophantine approximations operating ratio of the mixed frequency at the nonlinear frequency converter entering is made in order to localize the areas free from the order invalid combination frequencies Under the “calculation of the spectrum of nonlinear frequency conversion in the “near” area”, we mean the definition other next four combination frequencies coefficients, which form the “affected spots” near the working ratio q mixed frequencies f1 and f To determine near-area parameters of a nonlinear frequency converter, which is limited to four (or no more than four) closest combination frequencies with the exponent of less than allowed, means to look for the affected points formed by the combination frequencies As it is stated in6, the problem is reduced to finding the double Diophantine approximation with the help of Farey fractions Restriction on Farey number index is determined by the order of accounted combination frequencies The problem of finding the first Diophantine approximation ratio to the mixed frequencies q f1 f ( f1 d f ) in the class of Farey fractions R Q , using continued fraction mechanism is considered in6 and is an iterative process with the maximum number of iterations N ' log k ,where k – an admissible order of combination interferences The search algorithm of the first Diophantine approximation, which uses a mechanism of appropriate fractions, is the fastest and has no alternative6, is shown in Figure To find the second Diophantine approximations of choice when you want to search for the independence of the first and the second Diophantine approximation the most appropriate is the algorithm based on Farey-Cauchy theorem Based on the properties of Farey fractions, the maximum number of search iterations of the previous or subsequent Farey fractions is N " k There is a possibility of the implementation of the second search of Diophantine approximations using the results of Farey-Cauchy theorem, using direct and inverse algorithms 441 442 V Loginov et al / Procedia Computer Science 103 (2017) 439 – 446 Fig The first Diophantine approximation using continued fraction mechanism Let us consider the case of searching the nearest larger fraction According to the fundamental Farey-Cauchy theorem, determines the relationship of neighboring Farey fractions, states that if R j Q j M k , whereas Q j 1 is an integer such that Then R j 1 Q j 1 becomes M k fraction, immediately following R j Q j Representation of (1) and (2): k Qi Qi 1 d k (1) Ri Qi 1 { 1 mod Qi (2) Ri Qi 1 / Qi (3) Ri 1 Qi 1 d k ® ¯Qi 1 ! k Qi (4) Ri Qi 1 { mod Qi (5) Thus, the problem of finding the R j 1 Q j 1 fraction immediately following the R j Q j fraction, is to define the denominator Q j 1 by solving comparison (5) in the restrictions (4) To this we should transform comparison (5) into an equivalent equation in respect to two variables Q j 1 and mk 443 V Loginov et al / Procedia Computer Science 103 (2017) 439 – 446 Ri Qi 1 { Qi m x (6) Where mk - unknown integer multiplier Substituting Q j 1 boundary changes from (4) into (6) and solving the equation in respect to mk , we get the borders of mk mlow , mup change, which are determined on the basis of: ° mup ® °m ¯ low kRi Qi (7) mup Ri Block diagram of the direct algorithm that implements this process is shown in Figure For the R1 Q1 fraction expression (8) does not make sense, so in this case, taking into consideration Farey fraction properties, R2 Q2 k Structural reverse flowchart is shown in Figure In the case of search for the nearest lower value of a Farey fraction we use the feedback statement of FareyCauchy theorem and obtain algorithms similar to those in Figure and Using numerical experiment, it has been proved that the inverse algorithm, compared with direct one, has 38.2 % greater capacity12 Theoretical estimation of computational complexity, its real value over all the values of a number of Farey series used to calculate the first Diophantine approximation is S times more, but for the second Diophantine approximation–is S times more and does not depend on Farey series index12 With 0.9 probability, when calculating both Diophantine approximations the number of iterations required is twice less in comparison with the maximum theoretical estimates Stage Calculation optimal parameters of the frequency distribution in mixers for filtration all spurs given orders and the technical constraints on the implementation of the filter The basis for the introduction of classification of non-linear frequency mixer model is a division on criterion input frequency bands Consider three frequency mixer models on criterion conversion band: Mixer-subtracters – the primary use for signal processing system for conversion input wide spectrum from one frequency band into narrow spectrum to other frequency band Mixer-transponders – transferring a fixed frequency spectrum in other band Mixer-adders of input bands, the main application - in analog frequency synthesis systems Fig Direct algorithm of the second Diophantine approximation search Fig Reverse algorithm of the second Diophantine approximation search 444 V Loginov et al / Procedia Computer Science 103 (2017) 439 – 446 The main problem of the proposed models - determination of the frequency distribution in mixers Under the frequency distribution we understand the parameters of the relative nominal input and output frequency and their band The main constraints in determining the optimum parameters of the frequency distribution is filtration of spurs in “near” zone, which found in Stage To determine the nature of the frequency conversion ratio in mixer use q f1 f where f1 , f - fixed values of input frequencies, which are defined with respect to the lower and upper limits of the input frequency ranges in next form: C1 f1 f1n 'f1 (8) Coefficient C1 uniquely identifies the position of the frequency f1 relative to its own range ' f1 It is useful to describe the various frequency converters and is set by the developer or the previous mixer is determined within the frequency conversion system Subsequently, using the ratio C1 and the ratio of the mixed frequencies q can be divided analysis frequency conversion on two stages: the first stage is carried out a preliminary analysis “is afflicted” spurs in the zero approximation ( 'f1 0) , and the second stage to calculate optimal values 'f1norm and qopt Consider the ratio of input and output frequencies with f1 d f (i.e q d ) In this case, the filtering zone mixer displaced for summation frequencies (Figure 8) and for subtraction frequencies (Figure 9) The coordinates of the points’ filtration zone are presented in Table We are exploring the filtration zone of mixer in plane nomogram of spurs For this, we consider the boundary values of input and output frequencies (where f out Cout 1 M 1 M C1 , 'f1 norm M q ; S is the normalized input frequency range of mixer): Fig Mixer’s shema Fig Model filters For three basic models of the mixer was designed method of “pseudo conversion” line, which is based on noniteration mechanism for finding optimal parameters of frequency distribution - nominal frequency bands and their bandwidth Table Coordinates of the points’ filtration zone № point Abscissa qmin f1 f max qout max f out f2 qmax f1 max f max qout max f out f2 qmax f1 max f qout f out f2 qmin f1 f qout f out f2 Ordinate The search of optimal values qopt and 'f1norm get from pairwise solving the system of two equations: 445 V Loginov et al / Procedia Computer Science 103 (2017) 439 – 446 qj K fj ' f1norm S fj , qk K fk ' f1norm S fk (9) where, j 1, 4; k 2, define four points filtration zone of mixer in "near" area This determines the maximum number of the estimates in folowing conditions ( m 1, ; j 1, ; k 2,3 ): S fj S fk ' f1,norm m qm K K fi K fk norm fi ' f1, m ; (10) S fi , The optimal parameters 'f1norm and qopt determined from the following condition: 'f1norm ^ ` 'f1,norm ; qopt êơmin , Cmax ẳ m (11) The efficiency of the proposed method of “pseudo conversion” line is about five orders of magnitude in comparison with the use of brute force methods Stage Determining the values of spurs in mixer is based on the use of empirical methods8 The calculation the value of spurs on output mixer is a difficult task Levels of spurs depend on the type of nonlinearity, power level and the LO signal and the presence of the filter elements or specific interference suppression systems Tabular model it is one of the most common methods of calculating the levels of spurs8 The implementation of the method is to provide a two-dimensional array of empirical data on the level of spurs for known levels of input frequencies This model is the calculation of non-linear frequency conversion products gas one drawback - when converting down and up, in general case, levels of spurs are close, but different on level Thus, this model may be used for pre-simulation only To increase the accuracy of simulation mixer developed more complex empirical model - Global Mixer Model (GMM)9, based on the description of spurs behavior of mixers (asymmetric response in case up or down signal conversion) The model is based on the measurement of spurs using an automated hardware system & software to create a database and subsequent use during simulation with the possibility of interpolation measurements The proposed model evaluates and optimizes the behavior of the mixers in the systems when changing input power and the frequency in a wide range It will improve the accuracy of modeling nonlinear conversion mixer products GMM modeling nonlinear frequency conversion products behavior only in the frequency of the local oscillator signal frequency greater case Main feature of this model is taking into account asymmetry of nonlinear frequency conversion products in mixed frequencies proportions close to unity Other ratios of mixed frequencies are not considered Lack of GMM model is asymmetry (it has limited use in the mixed ratios and frequencies about at frequencies greater LO signal) To correct the shortcomings of the above models held their completion in the direction of increasing approximation points instead of the two used in the GMM model, at least five in the number of main areas nomogram combination frequencies Further complicating the asymmetric model nonlinear frequency conversion require information about input suppressing nonlinear frequency conversion products mixed at various ratios as the frequency for the case where the signal frequency is lower than the frequency of the local oscillator, and vice versa 446 V Loginov et al / Procedia Computer Science 103 (2017) 439 – 446 Fig Filtration zone of mixer for summation frequencies Fig Filtration zone of mixer for subtraction frequencies Five Point Empirical Mixer Model of the frequency converter (5PEMM) accounts the relationship nature and the local oscillator frequency signal It is designed file system for storing and description parameters, the main advantage is a combination of interference suppression account nonlinear frequency conversion in the frequency plan in all possible combinations of mixed frequency ratios Then determined levels of spurs in “near” area, taking into account the signal passes through intermediate frequency (IF) filter Basis of this model is to create an asymmetrical frequency mixer model So, it becomes possible to formulate more precise requirements for element base, filter and etc Using input parameters of mixer (which are input frequency range, the IF band and the minimum requirements on the parameters of the filter elements of mixer), we define the oscillator frequency band, the level of suppression of spurs in the “near” area and the parameters of the filter mixer Usage proposed three-step method to forecast the level of spurs and implementation on FPGA with a clock speed of GHz allows calculating level of spurs for 855 ns We take spurs up to 10 orders References Ashton K That ‘Internet of Things’ Thing RFID Journal – 22.06.2009 Cherednyak L Internet of Things Platform Open Systems SUBD Journal №7 – 26.09.2012 Joseph MitolaIII Doctor of Technology, Royal Institute of Technology, Sweden – May, 2000 Benzacar S Microwave product digest – July, 2014 LoginovV The nomogram: combination frequencies-algorithmic approach in view of the conversion on signal and local oscillator harmonics Radio engineering, – 2011 – p 61-66 Bidakarov A., Loginov V Analysis of the frequency distribution of the nonlinear frequency conversion systems algorithms Technical and environmental issues of river navigation, №269, VGAVT, N Novgorod – 1994 – p.67-74 Loginov V Models and non-iterative method nonlinear frequency conversion in the "near" area parameters optimizing Radio engineering and telecommunication systems No1, 2015– p.57-69 Manassevich V Frequency synthesizer (Theory and Design methodology) Svyaz – 1979 – p 384 – (edited by Galina A.) Liu J., Dunleavy L.P., Svensen T B European Microwave Conference – 2003 10 Sharapov Y Krilov G., Panteleev Y Converting the signal without the combination frequencies IPRZHR – 2001 – p 288 11 Galkin V Foftware-configurable radio fundmentials P: Hotline Telecom journal–2014–p 372 12 Loginov V.I Performance improvement of model spectrum algorithms of nonlinear frequencies transformation in “near” zone J Advances in modern Radioelectronics.No.9, 2015.p.62-69 ... parameters of the relative nominal input and output frequency and their band The main constraints in determining the optimum parameters of the frequency distribution is filtration of spurs in “near”... calculating level of spurs for 855 ns We take spurs up to 10 orders References Ashton K That ? ?Internet of Things? ?? Thing RFID Journal – 22.06.2009 Cherednyak L Internet of Things Platform Open Systems... shortcomings of the above models held their completion in the direction of increasing approximation points instead of the two used in the GMM model, at least five in the number of main areas nomogram