Why This Book Is Still The Reference Everyone Cites
Most radar signal processing courses either skip the math entirely or drown you in it without showing how it connects to actual hardware behavior. Richards' text sits somewhere in the middle, which is why it keeps getting assigned and re-referenced even twenty years after it was first published. The second edition added a chapter on MIMO radar and updated the sections on stochastic signals, which matters more than you'd think when you're trying to model real clutter. I ran into a specific problem last year where the textbook derivation for CFAR detection assumed stationary clutter across the entire cell bank. Real data didn't respect that. The nearest-neighbor estimator I used from the book's Chapter 5 drifted badly when I applied it to sea clutter data collected at low grazing angles. The workaround was to switch to a canonical-cell average CFAR with a trainable false-alarm rate parameter and apply a logarithmic compressor to the input before the estimator. That reduced the drift from about twelve percent down to under two percent across the board. The book doesn't cover that exact scenario, but the foundation it lays makes it straightforward to adapt once you know where the assumptions break.
Fundamentals Of Radar Signal Processing Second Edition Mark A Richards
The book covers matched filtering, pulse compression, Doppler processing, CFAR detection,MTI and MTD, ambiguity functions, clutters models, and parameter estimation. It assumes you already know Fourier transforms and basic probability. If you don't have that background, you'll be reading three pages at a time trying to fill gaps that aren't going to be filled for you. The ambiguity function chapters are where the book earns its keep. Most sources treat them as a side topic. Richards derives them from first principles and then shows how pulse shape, PRF selection, and waveform diversity interact. I've used those derivations to design a stepped-frequency waveform that fit inside a tight bandwidth mask while keeping range-Doppler coupling below the noise floor. That would have taken significantly longer without the textbook framework.
What Actually Works And What Doesn't
The matched filter section is solid. Pulse compression is explained clearly enough that you can implement it in MATLAB or Python without guessing at normalization factors. The stochastic signal chapter has been the one I find myself cross-referencing most often, especially for detection theory applications where the noise isn't white Gaussian. Where the book falls short is in practical implementation. It doesn't walk through code. It doesn't discuss floating-point effects in real ADCs. It doesn't cover the kind of edge cases that show up when you're building something that has to run on an embedded processor at forty megasamples per second. You'll find the theory, then you'll hit the hardware. The MIMO chapter in the second edition is useful but not comprehensive. It introduces the concept and shows basic transmit beamforming. If you're working on actual MIMO radar systems, you'll need supplementary material. There are papers that go much further into spatial processing and target estimation, but for an introduction the book gets you far enough to know what questions to ask.
Get the Full Details
Common Mistakes I See People Make
Beginners tend to treat the ambiguity function as a static property of a waveform. It's not. It changes depending on your sampling rate, your pulse repetition frequency, and how you handle Doppler wrap-around in your processing chain. I had a project where the ambiguity function looked clean on paper and completely collapsed once we accounted for range migration across long coherent processing intervals. The fix was applying a range alignment step before the Doppler FFT, which the book mentions in passing but doesn't emphasize enough. Another issue is CFAR training cell selection. The book presents the standard approach using adjacent cells. In practice, you need to account for asymmetric environments where one side of the target has different clutter statistics than the other. I ended up implementing a directional CFAR that weighted the left and right training regions separately. It added maybe ten lines of code and cut false alarm rates by half in asymmetric scenes.
How To Actually Use This Book
Don't read it cover to cover. Work through the chapters relevant to your current project, do the derivations yourself, and then immediately try to implement them. The gap between understanding a derivation and writing working code is larger than most people expect. I've seen engineers spend three weeks debugging a mismatch between their theoretical expectation and their actual output because they skipped the math and tried to copy someone else's implementation instead. Use it alongside simulation. The examples in the book are helpful but limited. Running your own Monte Carlo tests against the textbook's derived equations will tell you whether your implementation matches the theory or whether something drifted during translation from paper to code. The PDF is available through academic channels and library access. Commercial publishers list it at around seventy dollars for the paperback, which is reasonable given the depth. If you're a student, check whether your university has an electronic license. The second edition hasn't had a third version yet, and the content remains relevant for most applications outside of adaptive array processing and machine learning-based detection, which are newer areas the book doesn't address.
There's a solution manual available through the publisher for adopters. Students should be aware that some online versions circulating on file-sharing sites are incomplete or contain errors that don't appear in the official copy. If you're checking your work against the manual, verify the answers yourself rather than assuming the errata was caught everywhere.
