To be successful, efficient operation of networks apply the relevant methods of network and systems management, which require constant and reliable metrological provision and receipt of appropriate and effective estimates of the parameters measured random variables, processes or fields. To obtain estimates of the measured parameters can be used: sampling methods optimal estimate of random variables, recursive methods for estimating random variables or recursive evaluation methods of random processes. More constructive type ratings are recursive estimation procedure, which gives the current evaluation process in real time, and in the evaluation of the random variable recursive procedure tends asymptotically to the true value.
In the recursive methods for estimating the parameters evaluated included the process or the amount, and therefore the mathematical model. Therefore, actual scientific task is to develop a mathematical model of radio network LTE.
In LTE network every 0.5 ms to measure the characteristics of the channel, and every 40 ms reports are sent to the mean values of the measured parameters.
To provide access and distribution of resources subscribers can measure the following parameters: average power of the received pilot signal; quality of received pilot signals; ratio of the instantaneous data rate to an average speed; the ratio of signal power to interference power and noise power (SINR). The most obvious indicator is the SINR, as 1st and 2nd figures do not take into account the interference power in the channels, and the third figure is derived indicator of SINR. Therefore, measurements and assessments of quality indicators channels LTE networks rational use indicator SINR.
Measured SINR in uplink frequency band 1920 - 1980 MHz frequencies in adjacent 1920000 1920015 kHz kHz. According to the obtained samples calculated their mean values and variances. Built autocorrelation functions of sample data and cross-correlation function. Maximum values, these functions are at the value of the time shift is zero and slowly decreases with increasing time shift. Based on this schedule, we can conclude that they have an exponential form. According to Theorem J. Doob can be argued that the results obtained are Markov processes.
According to the correlation functions defined intervals correlation and cross-correlation intervals. To process one correlation interval is 21 count, which corresponds to the time interval 9.0006 ms. Process 2 correlation interval of 13 samples, which corresponds to the time interval 5.5718 ms. Cross-correlation interval is 7.5 counts. Consequently, during the time it is 3.2145 ms. Thus, it can be argued that the processes in the adjacent channels are statistically dependent.
Invited to the processes occurring in the channels to approximate a multivariate Markov model in the form of equations of state and observation equations. Calculated on the basis of the obtained samples of the average values of processes, variances and correlations intervals developed two-dimensional model equations of state and observation equations.