Table 1 clearly shows that all state-of-the-art sensors allow either for high resolution in com- bination with a narrow swath or can ensure wide-swath coverage with coarse geometric resolu- tion. As will be derived later in detail, this is caused by a limitation inherent to single-aperture SAR systems, which permits improvement in resolution only at the cost of a narrower swath and vice versa, as optimization of one of the two performance parameters inevitably results in a deg- radation of the other [18], [19]. Fig. 6 visualizes the relation between swath width and resolution, making clear the trade-off between both parameters and indicating the upper bound for achiev- able system performance by the gray-shaded area. It becomes obvious that SAR systems based on a single transmit and receive aperture, do not allow for high-resolution and wide-swath imag- ing at the same time.
Swath Width Level of Deta il Spotlight / Fine Resolution Stripmap ScanSAR / Wide Swath Syste m-in heren t Perform ance Lim it
Fig. 6. Relation between swath width and level of detail defined by the (inverse) geometric resolu- tion in conventional SAR systems. The gray-shaded area marks the limitation of achievable swath width and resolution as a given swath width cannot be imaged with an arbitrarily good resolution
and vice versa.
1.2 Motivation, Scope, and Structure of this Work
As introduced before, SAR is a well-proven imaging technique for the observation of the Earth. However, “conventional” SAR sensors, i.e. systems based on a single transmit and receive aperture, are inherently limited as they cannot provide high geometric resolution and wide-swath coverage at the same time (cf. [18] and Fig. 6). Nevertheless, especially in recent years, Earth remote sensing applications increasingly demand sensors that are capable of continuous global coverage and, in addition, provide detailed imagery. Thus, conventional SAR is not capable of meeting the rising demands of future remote sensing missions, and innovative concepts are needed to ensure the imaging of a wide swath with high resolution. The most promising of these concepts employ multiple receive apertures and are listed in chronological order by [20]-[60].2
2Chapter 3 provides a detailed overview on the different system concepts, methods and processing
This allows for the simultaneous reception by multiple channels, compared to conventional sys- tems with only a single receiving channel. The basic idea is, hence, to use the multiple receivers to gather additional information and to benefit from this information to overcome the above re- strictions of conventional SAR systems. In general, the multiple receivers can be either arranged in flight direction (“along-track”) [24], perpendicular to it (“cross-track”) [23], or in both dimen- sions [25]. Accordingly, processing methods for signals in elevation dimension are introduced [29]-[33], [38], [51] as well as along-track multi-channel data are considered [35], [39]-[41], [45], [48]-[50], or both dimensions are combined, using a joint approach [26], [28], [34], or by exploiting the SAR geometry [20]-[22].
The present work focuses on sensors with multiple apertures in the along-track direction. In such systems, the spatial separation of the data gathered for a single transmit pulse is determined by the spacing between the different receivers, while the distance between two subsequent pulses is defined by the pulse repetition frequency (PRF) and the sensor velocity. In consequence, the PRF affects the spatial distribution of the gathered data (“samples”) for a fixed antenna length in azimuth [36], [37]. When conventional SAR algorithms are applied to such data, system analysis reveals that the performance varies depending on the spatial sample distribution and is hence sensitive to unavoidable variations of operational parameters like the applied PRF. This means that conventional SAR processing will fail, and thus novel methods of processing the received multi-channel data are required [52], [53]. So far, the problem of developing an algorithm which is suited to ensure full system performance independently of the applied system parameters has not yet been answered satisfactorily.
Swath Width Lev el of D etail Conventio nal SAR Multi-Ch annel SA R High-Resolution Wide-Swath Multi-Channel SAR Multi-Channel SAR Multi-Aperture Processing Multi-Aperture Processing High-Resolution Wide-Swath Image Spotlight / Fine Resolution Stripmap ScanSAR / Wide Swath
Fig. 7. Multi-channel SAR system in combination with innovative digital multi-channel processing overcomes the limitation of conventional SAR and allows for high-resolution wide-swath imaging. The scope of this work is to introduce and analyze an innovative processing algorithm for azimuth data of multi-channel SAR systems in order to allow for high-resolution wide-swath imaging albeit with a varying PRF [42]-[44]. This work presents the derivation and verification of the novel algorithm as well as its analytical description allowing for the prediction of system performance [46], [47]. A detailed analysis in the frame of a system design example demon-
1.2 Motivation, Scope, and Structure of this Work 7
strates the capability to enable high-resolution wide-swath imagery and reveals the limitations of the algorithm which can be overcome by various innovative optimization strategies that are pre- sented and analyzed in the second part of the work [52]-[57]. In a final step, the algorithm is ex- tended to burst mode operation enabling a completely new class of SAR systems which is capa- ble of imaging so far unprecedented ultra-wide swaths with reasonable geometric resolution [58]-[60], [65].
In summary, the present work provides an innovative and very flexible “toolbox” with re- spect to multi-channel processing in azimuth. This includes the possibility of novel modes as well as introducing an exceptional flexibility and reconfigurability of SAR systems, thus ena- bling hitherto unprecedented performance in combination with a dramatically increased area of application. In consequence, the innovative processing algorithm in azimuth opens a completely new field of SAR operation, which allows for flexibly answering the needs of future remote sensing requirements. Especially if embedded in the frame of advanced imaging concepts [61]- [65], the presented azimuth processing methods show huge potential to become an indispensable component for the next decade of SAR.
This dissertation is organized as follows. It begins with an introduction to the basic princi- ples, relationships and constraints in conventional SAR systems to give a basic understanding of the topic and to stress the necessity for new SAR concepts (cf. Chapter 2). Then, Chapter 3 in- troduces the idea of multi-channel reception as a possible solution to overcome the system- inherent limitations of conventional approaches, followed by a detailed overview of the historical development of multi-channel SAR from its beginnings to state-of-the-art systems, concepts, and techniques. Focus is then turned to the azimuth dimension in Chapter 4, which presents a multi- channel SAR signal model. Furthermore, the properties of the spatial sampling in azimuth spe- cific to multi-channel systems are summarized in order to provide a better understanding of the signal processing analysis in azimuth. In Chapter 5, theory and the principle of the innovative digital beamforming algorithm in azimuth are presented, and the algorithm is incorporated in the classical space-time adaptive processing (STAP) framework. Then, the one-dimensional process- ing approach is embedded in the two-dimensional SAR processing scheme. Further, a detailed theoretical analysis is carried out showing how signal, ambiguities and noise are affected by the digital processing network. This provides analytic expressions to estimate the respective parame- ters in multi-channel SAR systems thus enabling a prediction of the system performance.
A proof-of-concept in Chapter 6 demonstrates the applicability and potential of the presented algorithm to multi-channel data obtained with the German Aerospace Center’s airborne sensors E-SAR and F-SAR.
Chapter 7 presents the design and performance analysis of an example system which enables the imaging of a swath width of 100 km with a resolution better than 1 m. Simulation results for various performance parameters in combination with a comparison to alternative azimuth proc- essing techniques demonstrate the great potential of the multi-channel reconstruction algorithm. In addition, the results allow for verification of the theory derived in Chapter 5. At the same time, the analysis illustrates the intricate connection between mission performance requirements and system parameters and highlights the resulting limitations.
In a next step, system optimization is discussed in Chapter 8 with the goal of overcoming the aforementioned limitations by improving the performance and increasing the flexibility of multi- channel SAR systems. At first, error sources are identified and innovative strategies are derived. Next, it is illustrated how an adaptation of the processed Doppler bandwidth and an adaptive management of the PRF in sparse array systems allow for an optimization of the system per- formance. Then, optimization concepts concerning the transmit side, such as pattern tapering on transmit and an innovative strategy based on the patented adaptive pulse-to-pulse shift of the transmit antenna phase center [56], [57], are explained. Afterwards, focus is turned to the receive side, and a new concept based on a cascaded structure of processing networks is introduced and analyzed. In this context, the theoretical examination of signal power, residual errors, ambigui- ties and scaling of the noise power is extended to the class of cascaded beamforming systems. The investigation of the cascaded beamforming technique is completed by a system performance analysis. In this framework, an analog representation of a cascaded network called “pre- beamshaping on receive” as well as a digital pre-processing approach are presented, and their performance is evaluated.
In a further step, the applicability of the multi-channel reconstruction algorithm to ScanSAR and TOPS-SAR is investigated and characteristic effects of multi-channel processing in burst mode systems are analyzed in Chapter 9. In this regard, a design example for such an innovative system concept is presented that enables the remote sensing of ultra-wide swaths of 400 km with a geometric resolution of 5 m.
This dissertation closes with a discussion containing an outlook on further issues like sparse array systems (cf. Chapter 10).