cfaed Seminar Series
cfaed Seminar Series
Prof. Josef Nossek , TU München
The Why and How of Coarse Quantization in Wireless Communications
, 10:00 - 11:30
Toepler-Bau, Room TOE 315 , Mommsenstraße 12 , 01169 Dresden
The challenges for the fifth genereation (5G) and beyond are enormous. One of the many goals of 5G is a 1000-
fold increase in network capacity, i.e. 1000 times more bits per second and area to be transmitted compared to 4G
networks. To achieve this dramatic improvement three symbioctic technological directions are indenpendently
exploited: small cells, millimeter-wave frequency bands and very large number of antennas at base stations (BS).
The last one commonly referred to a massive MIMO is expected to increase the power consumption because of
the large number of RF chains needed. Since energy efficiency is almost important as the increase in capacity
innovative solutions are desperately needed. To assess different approaches one has to identify the most power
hungry components in transmitter (Tx) and receiver (Rx). In Tx it turns out to be the high power transmit amplifier
(HPA), especially it has to be a linear one amplifying signals with high peak-to-average power ratio (PARR), while in
the Rx it is the analog-to-digital converter (ADC), especially if high resolution and high sampling rate are asked for.
One approach being researched quite extensively is the use of considerably fewer RF-chains than antenna
elements and implementing phase shifters (PSs) in the analog domain to combine digital signal processing and
analog phase shifting to so called hybrid beamforming. In the Rx this will drastically reduce the number of ADCs,
but will not help as much in the Tx. This is because of the number of HPAs should always be equal to the number
of antennas to avoid power loss in the analog PSs.
Although hybrid beamforming has got a lot of attention during the last few years, it is still not clear, whether there
will be a net improvement in energy efficiency compared to a full digital approach, especially if the latter one is
designed with low resolution ADCs (preferable 1bit!) in the Rx and with constant envelope signals at the HPAs in
the Tx. While the latter concept will lead to drastically reduced power consumption per RF-chain, the question is
how to mitigate the performance loss due to coarsly quantized signals.
This will be adressed in the second part: how the signal processing architecture has to be designed to provide the
same performance as a system with high resolution ADCs and highly linear HPAs. Obviously there is a price to be
paid for these energy efficient key components: the number of antennas – and the number of RF-chains – has to
be increased by approximately a factor of two. Although this may look quite substantial, it is expected to lead to a
considerably lower overall power consumption and, therefore to a higher energy efficiency.
The theoretical framework of the design of such a Constant Envelope-Coarsely-Quantized (CECQ) system will be
layed out, precording strategies will be described and a comparison of such a CECQ system with a standard linear
one in a multi user massive MIMO scenario will be carried out by numerical simulation. There are still quite a
number of open questions to be addressed to bring this promising concept to have practical impact.