Grasping coherency, synchronisation in MIMO systems

In modern communications systems, we are always looking for more: more data, more bandwidth, more coverage. Beyond using more processing power, more spectrum and more base stations, one option helping to provide greater capability is more antennas. Since the n revision of IEEE 802.11 was published in 2009, multiple-input multiple-output (MIMO) systems have become commonplace. As we look towards future wireless and cellular standards, however, MIMO will play an even more important role.

Applications of MIMO
Multi-antenna technology is important in a variety of microwave and communications applications. Direction finding relies on multiple inputs from a single transmitter, calculating the unique phase delay to each receiver to triangulate the location of the source of transmission. Beamforming uses multiple transmitters to output signals that will interfere constructively or destructively depending on the angle it is observed from, a technique used in active radar, jamming or focusing radiation in communications systems. Passive radar uses multiple receivers but no dedicated transmitter, instead relying on third-party transmitters in the environment and measuring the time of arrival difference between the signal arriving directly from the transmitter and arriving via reflection from the object. Finally, in cellular or connectivity systems, both multiple transmitters and multiple receivers are used to improve link quality and data rates.

The theory of operation
In MIMO systems, the effective signal-to-noise (SNR) ratio can be increased by transmitting unique bit streams with multiple transmit antennae in the same physical channel, known as spatial multiplexing. This technique takes a single data stream and multiplexes it into individual data streams, which differs from the traditional approach of growing data rates by simply using more spectrum. Prior to MIMO systems, a physical channel could be characterised by the amplitude, phase and frequency of the signal being transmitted. However, with multiple transmitters utilising the same portion of spectrum, the same models do not apply. For MIMO systems, multiple transmitters interfere with one another, causing the receiver to observe a signal that is a product of all transmitted signals. To make meaningful use of the signal, signal processing is used to reconstruct each of the transmitted streams and decode them individually. For this to be possible, the receiver must perform channel estimation, a technique that predicts characteristics of the transmitter and physical channel, including gain, phase and multi-path effects.

MIMO test challenges
Accurately testing MIMO transceivers brings challenges beyond those faced for single channel systems. In a multi-channel system, there are many sources of channel-to-channel phase variation, including uncorrelated ADC sample clock phase noise, uncorrelated LO phase noise, SNR and ADC quantisation noise.

The first area to consider is base band synchronisation. The data being transmitting begins life as a digital representation of the required signal, represented by in-phase and quadrature-phase information, I and Q. These two digital bit streams are typically passed into arbitrary waveform generators, which utilise a digital-to-analogue converter (DAC) to convert it into an analogue signal that can be mixed with a local oscillator (LO).

Ideally, the two ADCs should share one clock source, allowing them to be synchronised, but naturally there is a distribution delay in propagating the clock signal. This τ, the propagation delay, causes a phase skew between I and Q, which becomes a quadrature skew on the RF signal, leading to an increase in error vector magnitude (EVM). To reduce this, signal generators built on the PXI modular platform can share a 10MHz system clock, which can help reduce instrument-to-instrument skew to the order of 100 ps.

The importance of phase coherence
Once appropriate measures have been taken to synchronise base band I and Q, just as important is the synchronisation of the multiple channels that are in use. In this case, even sharing the 10MHz reference clock is not sufficient. In order to reduce channel-to-channel phase skew, local oscillators should be shared between multiple channels. With many instruments this is not necessarily straightforward, since the LO is not always exposed for sharing with other instruments, plus their size dictates the use of fairly long cable lengths, leading to longer propagation delays. Modular instrumentation has the advantage of being able to daisy-chain LOs from instrument to instrument, thus significantly reducing phase skew.