Edge computing has been recognized as a potential solution to enable intelligent manufacturing in the machining industry, especially for the small and medium-sized manufacturers. However, while various research studies have proposed their edge-based architectures for intelligent systems, there still exists a lack of practical and affordable technological plans that can be applied to complex machining process designs in actual production scenario. The objective of this research is to realize the tool condition monitoring (TCM) in machining by the edge computing based architecture for actual mass production. This study creatively proposes a calibration-based TCM system to monitor the cutting tool conditions in repetitive machining operations by comparing the characteristic signals generated by the reference cutting tools in the calibration procedure with the signal generated by the cutting tool in production through a concise similarity analysis, which can be easily integrated into typical cyber-psychical systems to realize the edge computing in a very efficient and flexible way. To validate the performance of the proposed architecture, a case study is demonstrated for tool wear monitoring of repetitive milling operations with a complex machining process design. Experimental validation has shown that the proposed edge-based TCM system can effectively monitor the tool wear progression which is in good agreement with actual wear measurements.