Nonlinearity in the RF sensing chain: single and multiple channel signal processing solutions

In today's digital world the radio frequency (RF) spectrum is becoming increasingly crowded. This poses a new interference challenge for conventional radars that operate within this RF spectrum. Any interference that is received by the radar is removed by filtering in the front end of the receiver. Historically this technique would be sufficient for screening out the majority interfering signals however this is no longer the case. The front-end of a radar receiver comprises of multiple amplifiers which are designed to have a highly-linear gain response. In a bid to reduce manufacturing costs cheaper amplifiers with considerably smaller linear regions are being used in the RF sensing chain. The linearity of the radar receiver has been sacrificed with potentially devastating effects for the sensor's overall performance. Before any detected interference is filtered out by the radar receiver the cheap amplifiers are forced into their nonlinear region. This is a regime the amplifiers have not been designed to operate in and therefore the signals become significantly distorted. While the outside interference is still being filtered out by the receiver its presence has generated unwanted effects that have not been removed.

Before any counter measures can be investigated the nonlinear distortion being generated must first be characterised. Radars are not designed to operate with their receivers in the nonlinear regime therefore very few attempts have previously been made to characterise their behaviour whilst in this state. This characterisation will be the first stage of the PhD and will involve combining hardware measurements with system modelling. Once the modulation effects are fully understood methods aimed at suppressing them can be studied. Most modern radars are designed to have multiple receive channels. The first mitigation technique designed in this PhD will aim to make use of these channels. Operating a receive channel with reduced gain would allow it to remain highly-linear even in the presence of interference. By configuring a few of the channels to be low gain an undistorted version of the signal could be captured and compared with the corrupted outputs. This signal comparison can be used to supress the overall distortion observed in the radar channels. This technique should be fairly mode independent but relies on deliberately reducing the performance of the radar. The second proposed mitigation technique aims to uses the receiver's nonlinear products to predict the original signal detected by the radar. It relies on having an extremely well characterised nonlinear receiver and exploits a unique property of the radar signal known as sparsity. This technique is expected to be mode specific and will be based around a lot of the concepts from compressive sensing.

It is predicted that a combination of the mitigation techniques will need to be employed to combat the distortion in a real scenario. While some of the mitigation ideas are being adapted from related fields there is very little published work surrounding this subject in the radar literature.