Like in the ecosystems of Nature, raw sensing is of little use unless we are also able to form higher-level interpretations of the collected data. How can we assess whether the sensed data is accurate? How can we tell whether a peculiar set of data arises from genuine conditions or is due to a faulty set of sensors? What is the normal operating condition of a digital sensor system? When is a deviation from normality to be interpreted as anomaly? This book explores the emerging area of sensor systems and applications from the particular perspective of anomaly detection. It gives the reader a head start on methods applicable to embedded sensor systems, showing the benefits of a range of computational approaches. After pinpointing the limitations of 'deterministic' anomaly detection, it becomes clear why the more promising approaches are those based on computational intelligence. The reader of this book will gain an in-depth understanding of anomaly detection in complex and unpredictable sensor systems, familiarizing with the most suitable machine learning techniques.