Signal processing antennas - Part 2

As with goniometry, the various imaging methods are based on an external environment that is not directly accessible (space, ocean, ground, organ, etc.). The aim is to provide a plausible image that can be used to locate various objects. Some environments are themselves sources of incoherent radiation (stellar radiation, thermal radiation, etc.). These give rise to passive imaging (radio astronomy, radiometry, passive sonar). This radiation is measured by sensors (receiving antennas) and processed using algorithms that take into account the information available a priori on the observed environment, as well as measurement noise.

Other media, such as active radar and sonar, are not the seat of spontaneous radiation. They must be illuminated by one or more transmitting antennas. For example, synthetic antennas, processed in , the field radiated by these antennas can be spatio-temporally coded (sometimes called colored radiation), to which polarimetric analysis can be added. One characteristic of active systems is that the field radiated on emission is coherent, whereas the field backscattered by the medium analyzed is generally at least partially decorrelated. This decorrelation is particularly noticeable in the presence of multiple signals and jammers. It may also be due to the propagation medium or the mobility of the receivers.

A first simple example of a method for imaging incoherent sources will be given with multiplicative arrays, which have a wide range of applications (harbour radar, radio astronomy...).

We then present the important Van Cittert and Zernicke theorem. It relates the (unknown) angular distribution of external sources to the spatial coherence function of the observable field. Sensors associated with correlators provide a spatial sampling of this function: the covariance matrix, which plays a central role in all antenna processing.

In all cases, the quality of the image obtained is limited by the separating power or resolution limit of the instrument used (antennas and associated processing).

Classically, the angular resolution of an antenna is limited by its dimensions, which in turn limit the fineness of the brush it can radiate. High-resolution methods show how, in certain cases, it is possible to overcome this limit: we will present BURG's method, known as the Maximum Entropy Method (MEM). This method is often equivalent to Autoregressive (AR) spectral analysis methods. Other methods, based on analysis of the eigenelements of the covariance matrix, will also be presented, illustrated by the best-known: the "MUSIC" method.

We'll go on to study self-adaptive antennas, which perform a kind of spatial filtering, favoring useful...