ПРИМЕНЕНИЕ МЕТОДОВ НЕЧЕТКОЙ КЛАСТЕРИЗАЦИИ ДЛЯ ЭФФЕКТИВНОГО УПРАВЛЛЕНИЯРЕСУРСАМИСОТОВОЙ СВЯЗИ

Стрельникова Л.В.¹, Зотов К.Н.², Кузнецов И.В.³, Жданов Р.Р.⁴

¹Студент, ²Кандидат технических наук, старший преподаватель, Уфимский государственный авиационный технический университет, ³Доктор технических наук, доцент, Уфимский государственный авиационный технический университет, ⁴Кандидат технических наук, доцент, Уфимский государственный авиационный технический университет

ПРИМЕНЕНИЕ МЕТОДОВ НЕЧЕТКОЙ КЛАСТЕРИЗАЦИИ ДЛЯ ЭФФЕКТИВНОГО УПРАВЛЛЕНИЯРЕСУРСАМИСОТОВОЙ СВЯЗИ

Аннотация

В рамках дано работы описывается необходимость кластеризации абонентов. Представлен оптимальный алгоритм кластеризации и результаты решения задачи по кластеризации большого числа абонентов с его применением.

Ключевые слова: алгоритм нечеткой кластеризации, Fuzzy C-Means, FCM  алгоритм, узел спроса.

Strelnikva L.V.¹, Zotov K.N.², Kuznetzov I.V. ³,  Zhdanov R.R.⁴

¹Student, ²Candidate of Technical Sciences, senior Lecturer, Ufa State Aviation Technical University, ³ PhD, Technical Sciences, Associate Professor, Ufa State Aviation Technical University, ⁴Candidate of Technical Sciences, Associate Professor, Ufa State Aviation Technical University

THE APPLICATION OF FUZZY CLUSTERING METHOD FOR EFFICIENT MANEGEMENT OF CELLULAR RESOURCES

Abstract

Within the scope of this work we the necessity of clustering for cellular subscribers is given.  Optimal clustering algorithm and the way it solves the problem of big amount of subscribers are described in this work.

Keywords: fuzzy clustering algorithm , Fuzzy C-Means, FCM algorithm , the node demand.

The radio resources are subject to operational control because when subscribers move it can cause occasional congestions in mobile communication systems.  The resources from less loaded parts can be used to solve this problem. [1].

Positioning function in MS mobile operator networks is an important problem of modern science.[2]

First step to provide efficient management of radio resources is the location of mobile stations (MS) with accuracy which help to find areas with high concentration of subscribers and identify nodes of demand inside each cluster.

The set of fuzzy areas of possible appearance of subscribers is to be clustered. The most appropriate fuzzy clustering algorithm for that is FCM.

The first data processing by change theory brings information about sufficient quantity of clusters for default variety of subscribers.[3] As a result we can clearly identify the nodes of demand on a region map.

The FCM algorithm’s features are:

Possibility to set the centers of future clusters in space by prior information about concentration of subscribers. Derivation of clustered areas of subscribers with assumptions about membership in particular cluster. Possibility to set quantity of clusters, and to derive the centre of mass of clusters.

The clustering algorithm based on FCM consists of the following steps:

Step 1. Location the centers (the point used for enumeration is not the center of cluster)

Step 2. Determination of necessary quantity of clusters. (from data obtained after the change theory application)

Step 3. Operation of algorithm.  Minimization of sum of weighted distances where :

q – fixed parameter set before iterations.  It is assumed for test set K of input vectors dk and N isolated clusters cj that every dk belongs to every cj with membership µjk to interval [0,1] where j – cluster number, k – input vector number, ||   || - matrix norm (Euclidean norm) and ε – predetermined accuracy level.

The following terms of normalization for:

Weighted center of gravity is following: and Step 4. The k-means is considered completed if the following condition is fulfilled:

Step 5. Determination of the center of mass of obtained fuzzy figures in the form of final coordinates.

The fuzzy clustering algorithm can be used as an adjusting tool. The base stations of cellular operator should be placed in identified nodes of demand.

Thus, clusters with the centers of masses were defined in final sets of subscribers of the cellular operator. These centers of mass of the obtained clusters provide information about points of highest congestions which are nodes of demand in the mobile communication systems.

References

1. Sultanov A.H., Kuznecov I.V., Kamalov A.Je., Ob odnom metode prognoza optimal'noj zony radiopokrytija seti mobil'noj svjazi. – Vestnik UGATU, 2010 god, 62-67 str.
2. Titov, E. V. Opredelenie dopustimogo vremeni prebyvanija v zone vlijanija jelektromagnitnyh izluchenij [Tekst] / E. V. Titov // Vestnik AGAU. – Barnaul, 2014. – № 3 (113). – S. 49 - 54.
3. Zotov K.N., Razrabotka algoritma povyshenija tochnosti pozicionirovanija mobil'nyh stancij na osnove rascheta staticheskih parametrov jelektromagnitnogo polja v neodnorodnoj srede. – Vestnik UGATU, T.17, №2(55), Ufa, 2013. – 14 – 19 str.