Signal Recognition And Attribute Reduction Based On Fuzzy Preference Rough Sets

With the rapid development of the information age, the radio spectrum resources hasbecome an important resource for widespread use of human society, which has put forwardthe new requirements for the radio signal monitoring. The recognition of modulation types isone of the important tasks of signal monitoring, and characteristic signal presentationcontinuity, high dimensional, nonlinear characteristics caused great distress to our recognition.Application of rough set model to deal with complex data has many advantages, so applying itto reduction of signal characteristics, has important practical significance to the recognition ofmodulation types.As claimed in many studies, fuzzy sets and rough sets are complimentary rather thancomplementary in dealing with uncertainty. Fuzzy preference rough set model is a fuzzyformal dominance rough set. It is a generalized fuzzy rough set model which is by replacingdominance relations with fuzzy preference relations. This paper is mainly research attributereduction and recognize signal on C band radio signals based on the preference rough setmodel. The main achievements of the thesis are summarized as follows:(1) According to the structural objects of multiple criteria decision information systembetween the fuzzy preference relation, this paper gives a new simpler criterion function togenerate the fuzzy preference relation, proved that the new fuzzy preference relation and f therelation [12] is consistent, namely does not change the object’s preference relation.(2) Based on fuzzy preference rough set model, this paper defines a new fuzzyapproximation quality, and presents the calculation method of fuzzy preferenceapproximation and fuzzy preference approximation quality. And then defines the attributereduction of information system, prove the reduction that is the minimal set of attributeswhich kept lower approximation unchanged. Finally, we propose a heuristic algorithm forcomputing reduction by the importance of attributes.(3) The radio signal characteristics are analyzed. We did the experiment to contrast theaccurate rate of signal recognition by individual attribute and combined attributes. Andthan the flowchart of signal recognition is given. Finally points out the existing problemsin the current signal recognition.(4) Take the C-band radio signal as the research background. The attributes of radiosignal are turned into criteria, construct a multi-criteria decision information system anddelete the unnecessary attributes. Evaluation values belonging to each type are aggregated bythe OWA. Considering these evaluation values, if there is only one value which approximates1, at the same time the other evaluation values are close to0, then the recognition of the signal is the type of whose value is the highest. Otherwise, the recognition of the signal is a new typewhich is not one of the categories.