For millimetre wave communication we need site specific information about the particular environment for smooth communication between BS and UE (User Equipment) because propagation distance of millimetre wave in atmosphere is limited without beamforming or directional transmission of signal.
Figure: Beam training with analog architecture
Channel estimation is to acquire site specific channel model of a particular environment. Millimetre wave cannot travel much distance in atmosphere due to its severe pathloss and high refraction properties. It also experiences high penetration loss. So, we need to equip MIMO system to avail beamforming. And it is also needed to deploy large antenna arrays in MIMO system to form narrower and stronger beam to fulfil the link margin gain at receiver side.
So, we need to train the beam to find the best beamforming vector at transmitter side and best combing vector pair at receiver side to maximize the signal to noise ratio. The beam training can be realized by analog beamformer but the limitation of analog beamforming is that only one data stream with same amplitude and different phases is possible. That cannot provide high data rates as required. That’s why we need to focus on MIMO hybrid architecture. In that architecture analog beam steering is realized by large antenna arrays and baseband precoding is realized by low dimensional digital precoder. It also helps us to maintain low overhead in network.
Especially, for vehicular communication channel varies fast, so, channel estimation plays a crucial role to find the best communicating beam from BS to find the best signal to noise ratio from beam steering.