Multipath Interference and Diversity Switching
A significant challenge for engineers designing a wireless microphone system is addressing multipath interference. The Audio Ltd A10 digital wireless microphone system devotes significant signal processing resources to achieving robust performance in the presence of both short range and long range multipath. This note briefly explains what multipath is, why it is a problem, and some of the techniques used to mitigate it.
What is Multipath Interference?
Multipath interference occurs when an RF signal from a transmitter arrives at a receiver via two or more routes. Typically there is a direct path plus a number of indirect paths caused by reflections. Walls, people, set pieces, and other objects in a room can cause reflections indoor. Nearby buildings and vehicles can cause reflections outdoors.
In general, the total length of each signal path will be different. Therefore, the time delay associated with each path will also be different, which in turn causes the phase of each received signal to be different. If signals arrive approximately in phase, received signal power increases. If signals arrive out of phase, the signal is attenuated.
Flat and Frequency Selective Fading
As a transmitter or receiver moves, the relative phase of the multiple signals changes, causing the total received power to fluctuate. This is known as fading. As well as being affected by the path differences, the signal phases are also affected by the transmitted frequency. If the path difference between the various signals is short, the entire signal tends to be affected by this fading. This is sometimes referred to as flat fading. But for longer path differences, the phases of the received signals vary across the signal bandwidth. The signals can therefore interfere constructively or destructively at different frequencies in this bandwidth. The effect is to induce a slope in the system frequency response, or even to create multiple peaks or notches in the received signal spectrum. This is known as frequency selective fading.
The path length difference at which frequency selective fading starts to become significant depends on the signal bandwidth. With a path length of c/4B (where c is the speed of light and B is the RF signal bandwidth), there is the possibility of a severe slope across the band. For a typical wireless microphone system with 200 KHz RF bandwidth this corresponds to 375 meters. However, path lengths of reflections can be several multiples of the direct path length (if there is one), so frequency selective effects can start to be seen at much shorter operating range – well under 100 meters.
In practice, for typical indoor operation, the primary effect is flat fading. Outdoors, particularly in an environment like a sports stadium or a city street, both types of fading may be encountered.
How Multipath Affects Wireless Microphones
Considering frequency selective fading first, the effect on an analogue FM system is to introduce audio distortion or, in severe cases, signal drop-outs. A simplistic digital system is also impaired by this fading—prone to digital errors and signal drop-outs as the receiver or transmitter moves.
However, it is possible to equalise the channel frequency response and apply error correction techniques such that a digital system has excellent immunity from frequency selective fading. The advanced digital signal processing used in the Audio Ltd A10 Digital Wireless System benefits from the additional signals present in multipath reflections.
Flat fading is in some ways worse than frequency selective fading. It can effectively cause a complete loss of signal—no amount of signal processing can correct for this.
The most common way to manage the effects of flat fading is to use a diversity antenna topology, where more than one antenna receives signal. In sophisticated systems such as the A10 digital system, its advanced diversity system can help with both flat fading and frequency selective fading.
As a transmitter or receiver is moved, the phase of the received signals vary. The probability that the signals received at an antenna at a particular location sum and cancel to nearly zero is quite small. But in a typical room there may be numerous “dead spots” with severely attenuated reception.
If two antennas are placed at different locations, the chances that both are at a signal null is much lower, as long as the antennas are far enough apart to effectively de-correlate the signals. Ideally, the antennas should be as far apart as possible, though a minimum distance of half the wavelength of the transmitted signal frequency is generally effective. With a transmitting frequency of 500 MHz half a wavelength is roughly 30 centimeters.
Another way of de-correlating signals is to orientate the receiving antennas at 90 degrees to each other. The two antennas receive different signal polarisations (reflections tend to scramble signal polarisation randomly). In this case the antennas can be co-located.
Implementing Antenna Diversity
How do we make use of two or more antennas in a wireless system? In principle, multiple antennas could be used at the transmitter, or receiver, or both. It is typical for portable wireless microphone systems to use a single antenna on the transmitter and diversity antennas at the receiver because of the need to minimise transmitter size and power dissipation.
A simplistic diversity implementation is to use an electronic switch to connect one of two antennas to a single receiver. The receiver monitors the signal quality, and when signal at one antenna falls below some threshold, it switches to the other antenna. There are two disadvantages to this simple approach. Firstly, the receiver has no knowledge of the signal quality from the unused antenna; it may make a false choice and switch to an antenna with worse signal quality. Secondly, particularly in an analogue system, it is difficult to make the switching between the two signals completely inaudible.
A better, and widely used, approach is to have two complete receiver chains. This is sometimes referred to as ‘true diversity’. As with the simple receiver, this can be used in an analogue or a digital system.
In the case of the analogue system, both receiver chains make some measure of signal quality (often measuring signal strength). At a point in the demodulation process, usually immediately after the FM demodulator, the output from the better signal chain is selected and sent to the final output. A slight improvement on this is to perform a cross-fade between the audio from the two chains. The signals are weighted according to the signal quality.
Diversity Switching in Digital Systems
In a digital system, the diversity switching is usually implemented on data streams before audio decoding. At the expense of a small delay in the data streams, it is possible to make this switching completely seamless, and effectively anticipate the loss of signal of one stream or the other before it happens. This is sometimes known as selection diversity, and is highly effective.
However, digital receivers can use additional strategies to improve performance. The A10 Digital Wireless System carries out two additional processing steps, 1) it calculates a weighted sum of the two signals and, 2) the combining ratio varies with frequency.
Even if the signal from one of the antennas is below the signal / noise decoding threshold of the data demodulator, it still contains some useful information. Instead of performing a hard switch between the data streams from the two receivers, a weighted sum of the two signals is calculated. The weighting depends on an estimate of the channel signal / noise ratio (the approach is known as maximal ratio combining). This gives theoretically optimum reception in a non-frequency selective (flat) channel.
However, to improve reception given a frequency selective channel, a further step is possible. Instead of having a single combining ratio for the whole channel, the combining ratio varies with frequency across the channel to take account of the frequency selective fading. This combining is then theoretically optimum for both frequency selective and non-frequency selective channels.
What is the practical consequence of this? It turns out that this approach results in a 3 dB improvement in receiver sensitivity compared to selection diversity even without any multipath. Given the presence of flat fading or frequency selective fading, the improvement can be substantially greater, depending on the severity and nature of the fading.
Managing multipath interference is an essential feature of any wireless microphone system. The A10 system from Audio Ltd uses several advanced signal processing techniques to maximize robustness under the most difficult reception conditions.