In this paper we present the contactless method of analyzing the cardiac activity of a person based on recording and analyzing a ballistic cardiogram signal. A measuring device for registration of microscopic movements of the body uses a piezoelectric sensor of high sensitivity. Due to sensor’s high sensitivity, the level of background noise is higher than the signal level, so mathematical methods are used for noise reduction. Butterworth filter is used to extract cardiac signal. This approach is more computationally efficient compared to machine learning-based methods, and can be implemented on an edge computing node to which several sensors are connected. The quality of the signal obtained after filtration allows us to detect cardiac cycles. The algorithm used for detection of heartbeats proposed in this paper is also computationally simple enough to be implemented at the edge node. After preprocessing described above data is transmitted to the datacenter (cloud).