Wireless channel load stress analysis using FPGAs at the edge
Ganewattha, Chanaka (2019-07-04)
Ganewattha, Chanaka
C. Ganewattha
04.07.2019
© 2019 Chanaka Ganewattha. Tämä Kohde on tekijänoikeuden ja/tai lähioikeuksien suojaama. Voit käyttää Kohdetta käyttöösi sovellettavan tekijänoikeutta ja lähioikeuksia koskevan lainsäädännön sallimilla tavoilla. Muunlaista käyttöä varten tarvitset oikeudenhaltijoiden luvan.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:oulu-201910162972
https://urn.fi/URN:NBN:fi:oulu-201910162972
Tiivistelmä
One of the key usage scenarios of fifth generation (5G) and beyond networks is to provide mission critical, ultra-reliable and low latency communications (URLLC) targeting specific set of applications where low latency and highly reliable wireless links are of utmost importance. 5G and beyond applications that require URLLC links include industry automation, artificial intelligence based technological solutions, vehicle to vehicle communication and robotics enabled medical solutions. URLLC applications using wireless connectivity require that resource utilization, such as wireless channel utilization, does not exceed the levels above which performance can degrade. Real-time radio frequency (RF) data analytics at the wireless network edge can help to design proactive resource allocation solutions that can allocate more radio resources when a particular resource is forecasted to be under stress. Typically, real-time RF data analytics can require processing of hundreds of millions of streaming samples per second and hardware accelerated modules (such as FPGAs) are very well-suited for such processing tasks. We propose FPGA-accelerated real-time data analytics based resource stress forecasting method in this thesis. The proposed method is low in complexity and performs forecasting in real-time. We show its implementation on an FPGA of Xilinx Zynq-7000 series System on Chip (SoC) board using Vivado, Vivado HLS, SDK and MATLAB tools. The proposed method uses quantile estimation and can be used for forecasting a variety of resource utilization scenarios. As an example, in our thesis, we focus on forecasting stress in wireless channel utilization. We test the implemented algorithm with real wireless channel utilization data representing block maxima series. We compare the results from the implemented method against the results from a theoretical method where the generalized extreme value (GEV) theory is used to make forecasts on the considered block maxima data. We show that with high accuracy and low latency, the proposed algorithm can perform the forecasting of channel utilization stress.
Kokoelmat
- Avoin saatavuus [31928]