Browsing by Author "Rantanen, Essi"

Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    Kvasi-Monte Carlo
    (2018-09-28) Rantanen, Essi; Hakula, Harri; Perustieteiden korkeakoulu; Hakula, Harri
  • Thumbnail Image
    Item
    Study of the Statistical Properties of SRS and DMRS for Machine Learning in 5G
    (2021-12-14) Rantanen, Essi; Medeiros, Luiz; Perustieteiden korkeakoulu; Wichman, Risto
    In today’s telecommunications world latency, data rates and reliability are crucial components in the quest to gain a market edge. 5G New Radio has been introduced as the successor of 4G. In order to achieve the potential of 5G new strategies and technologies must be explored. This thesis focuses on studying the statistical properties of two uplink reference signals: sounding reference signal (SRS) and demodulation reference signal (DMRS), in the context of machine learning in 5G. In this study, the goal was to map the fundamental similarities and differences between the two signal types, in an effort to determine whether a common machine learning algorithm could receive both signals in its input feature set. It was done by examining SRSs and DMRSs before and after channels with various mathematical tools. The results demonstrated the importance of Zadoff-Chu (ZC) sequence in the context of feature engineering, as the signals based on ZC experienced low autocorrelation even after they had been reshaped in a form suitable for ML. In addition, it was also possible to show that SRSs and DMRSs hold similar statistics. Finally, theunique features contained therein also indicated that it is theoretically possible for one ML algorithm to use both signals as input.