COMPARISON OF OPTIMIZATION STRATEGIES AND ESTIMATION TECHNIQUES FOR RADIO NETWORK PLANNING AND OPTIMIZATION PROBLEMS

Background. Radio network planning is one of the main phases cellular lifecycle, as it determines capital and operating costs allows system performance evaluation at any given time. An accurate comprehensive analysis existing statistics necessary for proper cell during expansion. These are collected throughout life cycle usually have certain imperfections (heterogeneity statistics, which different densities in parts search space, up to presence significant voids, etc.) The describing functioning radio can be represented a black box because its internal processes too complex defined by mathematical functions. This need use appropriate tools.
 Objective. purpose paper create toolkit that finding relationships between parameters define target values will help build an effective plan terms creation operation. tools should able work efficiently using minimum set available statistical data, well taking into account their imperfections.
 Methods. Mathematical estimation optimization methods used, namely Ordinary Least Squares, Ridge Regression, Lasso, Elastic-net, LARS lasso, Bayesian Automatic Relevance Determination, Stochastic gradient descent, Theil-Sen estimator, Huber Quantile regression, Polynomial regression. We consider 12 combination with two strategies. Additionally, method partial space number configurations considered.
 Results. A software package Python programming language has been created, contains practical implementation all considered methods, evaluating arbitrary (benchmark) visualizing results. best Squares optimal configuration 4G maximize download speed. To obtain satisfactory results, enough 25 initial 250 estimated points - larger not significantly increase prediction accuracy.
 Conclusions. results indicate possibility created tasks. Further research aimed expanding package's functionality considering additional strategies.
 Keywords: planning; optimization; networks; techniques; strategies; big data; 2G; 3G; 4G; CQI; capacity.