A contribution by Dr. Goran Panic, Scientist (System Architectures / Wireless Broadband Communication Systems) and coordinator for Hub 2 in the Green ICT project at Leibniz Institute for High Performance Microelectronics (IHP).
In this tech blog, we present the activities of the IHP in Hub 2 of the Green ICT project related to the construction of a measurement hub to record all parameters of wireless communication systems relevant to the life cycle assessment. This includes the construction and automation of an anechoic chamber that is optimized for the characterization and analysis of RF communication devices that operate in the mm-wave and sub-THz spectrum. To validate the setup, we carried out the power and performance analysis of a 60 GHz 4x4 LoS MIMO system based on commercial radio modules.
As part of the project activities, the IHP equipped a 7 m x 4 m x 2 m anechoic chamber with a new positioning system. This consists of a fixed rotation positioner and a linear translation positioner with rotation function. A new control unit with four control outputs for the positioners was installed and software was developed to automate the measurements. In addition, various measurement setups for RF and antenna measurements were defined and the necessary measurement equipment was selected. The selected equipment for RF component characterization and spectral analysis includes a spectrum analyzer and an oscilloscope that operate at frequencies up to 67 GHz, but can also perform measurements in the range up to 300 GHz with appropriate extender modules. A range of waveform generators are made available for modulated signal generation and hardware-in-the-loop measurements. The power consumption measurements are carried out by programmable multi-channel power supplies. These were integrated into the automated environment, enabling remote-controlled measurements and data recording. To verify the installed test setup and established procedures, measurements were performed on a 60 GHz 4x4 LoS MIMO link setup (Fig. 1).
The installation was based on commercial 60 GHz radio modules from Analog Devices. The used RF frontends are 802.11ad compliant having an RF bandwidth of 2 GHz. They feature a programmable IF and RF gain control and are equipped with a 20 dBi horn antenna. Before setting up the MIMO link, the Tx output power of each individual module was characterized in the anechoic chamber. In addition, the power consumption was also measured for each gain setting (Fig. 2). The antenna radiation pattern was not measured as it was provided by the module manufacturer.
After characterizing the gain and performance of individual modules, a 60 GHz 4x4 LoS MIMO link setup was built in the anechoic chamber in a hardware-in-the-loop configuration (Fig. 3). In addition to the two 4-channel 60 GHz transceiver arrays, the setup included the necessary signal generation and recording equipment as well as the multi-channel power sources with integrated power consumption recording functionality. The transmission parameters were optimized at 5 m distance and the transmitted signal was QPSK modulated. The aim was to measure the signal quality (rmsEVM) and bit error rate (BER) for different gain settings and to associate them with the modules power consumption (Fig. 4). The measurements were performed for SISO, 2x2 MIMO and 4x4 MIMO links. The used signal bandwidth per QPSK stream was 1.76 GHz. Considering the spectral efficiency of a QPSK signal of 2 bit/s/Hz, the total data throughput per single stream was 3.52 Gbit/s. Accordingly, the aggregated throughput for 2x2 MIMO was 7.04 Gbit/s and for 4x4 MIMO 14.08 Gbit/s. The measurements showed that for the specified range and signal parameters, the bit error rate of the SISO link is less than 1E-3 at the gain setting of “3” (-6 dBm). The same bit error rate was achieved for 2x2 MIMO and 4x4 MIMO at the gain setting of “2” (-8 dBm). In numbers, the measured dynamic power consumption of the 60 GHz front-end in the SISO link was 1.805 W, which corresponds to a power consumption weighted by the data throughput of 0.513 W/Gbit/s. The measured dynamic power consumption did not include the static power consumption of approx. 1.5 W, which refers to the constant power consumption of the carrier board. However, when the power consumption is normalized to the data throughput, as expected it remains more or less constant and only increases slightly with a larger number of streams.
The analysis has shown that for the measured MIMO link setup, the best power saving strategy is to choose a higher modulation order to minimize the number of required MIMO streams. This is explained by the fact that increasing the gain to support higher order modulations has relatively small impact on the dynamic power consumption in the measured frontends. However, for a better assessment, it is necessary to include the impact of signal processing on the total power consumption, since higher order modulations are also expected to increase the power consumption of signal processing.
