Abstract:
As 5G continues to gain more momentum around the world, there are still challenges that need to be addressed in order to capitalize fully on the benefits of the proposed architectures and technologies, that include small cells, advanced OFDM, beamforming, massive MIMO, and millimeter wave. A particular, challenge is 5G channel estimation due to the large and high frequency range involved. In this paper, conventional channel estimation methods such as least square, Minimum Mean Square Error, blind and semi-blind are investigated. Moreover, the application of machine and deep learning in channel estimation has been discussed. We then use the IBM Watson machine learning service for channel estimation using the DeepMIMO dataset which yielded promising results.
Description:
The fifth generation of mobile communication (5G) was created to accommodate the exponential increase in data and the need for reliable communications for emerging technologies. The application of IoT in industries, smart cities, connected heath care etc., has contributed to this exponential growth of data (Big data) and due to the nature of the devices used, where high data speed communication with minimal latency is required. The 5G is expected to increase data rate to about 10Gbit/s second or more, reduce the latency by 10 fold, and lower the power consumption. As a result, the millimeter wave band has been considered as the suitable spectrum to deliver on this required results. The mmWave spectrum ranges between 30GHz and 300GHz, with wavelengths between 1mm and 10mm.
