The most interesting feature of negative selection algorithms is ability for detecting novel, never met anomalies. This is especially important in security systems like intrusion detection, spam, virus detection, etc. However, the main problem is scalability which occurs for both: binary and real-valued representation. This paper describes a hardware implementations of the process of generating b-detectors which allows for a fast generation the receptors as well as a very fast recognition of anomalies in high-dimensional datasets.