Identifying the Cyber-Incidents in Additive Manufacturing Systems via Multimedia Signals

Abstract

Additive Manufacturing (AM) is an emerging manufacturing technology that plays a growing role in both industrial and consumer settings. However, security concerns of the AM have been raised among researchers. In this paper, we present an online detection mechanism for the malicious attempts on AM system, which taps into both audios and videos collected during the actual printing process. For audio signals, we propose to monitor the characteristics or patterns in the spectrogram via the Wasserstain metric. For video signals, we present a path reconstruction method which effectively monitors the motion of the printer extruder. We then show the effectiveness of our methods in a case study using Ender 3D printer, where the cyber-incidence of modifying the internal fill density can be easily identified in an online manner.

Online detection of cyber‐incidents in additive manufacturing systems via analyzing multimedia signals

“…There are also works using the reconstructed path for attack identification through using acoustic sensors or gyroscopic sensors 4, 17 . In our previous work, 16 we briefly discussed the possibility of detecting attacks via path reconstruction in AM using video signals. Nevertheless, none of them propose a method to quantify the divergence between the reconstructed path from a test process and reference path.…”

“…11 There are also growing numbers of research utilizing the side-channel analysis, that is, monitoring the physical signals emitted during the manufacturing process to detect cyber-attacks. [12][13][14][15][16] In a recent study, 14 cyber-attacks that target the modification of design files can be successfully detected with learning methods in analyzing the captured images during manufacturing. However, this method is hard to generalize to other printers as taking images of the print would be particularly hard when the extruder of the printer is of close distance to the printing surface.…”

“…To tackle the issue of covert attacks, one of the most difficult to be detected, a data‐driven framework that taps into the degradation process of the systems, was developed 11 . There are also growing numbers of research utilizing the side‐channel analysis, that is, monitoring the physical signals emitted during the manufacturing process to detect cyber‐attacks 12–16 …”

A tensor‐based hierarchical process monitoring approach for anomaly detection in additive manufacturing

“…A rank one tensor can be produced by extending the concept of vector outer product to tensor product for 𝑁 vectors. For example, an 𝑁-way tensor of rank one can be represented by  = 𝒂 (1) •𝒂 (2) • ⋯ ⋅ 𝒂 (𝑁) where the 𝒂 (𝑘) is the a vector of dimension 𝐼 𝑘 . The rank of a tensor  is defined as the smallest number of rank-one tensors whose sum can generate .…”

“…Additive manufacturing (AM), also known as 3D printing, has seen increasing number of applications in the era of industry 4.0 due to its ability of creating new and complex shapes that overcomes the limits of traditional manufacturing processes and reshapes the whole value chain. 1,2 AM technologies also contribute to the digital and twin transition of the advanced manufacturing industry, thanks to new levels of data availability throughout the entire duration of the process. Such big data can be acquired by means of a variety of different sensors installed on AM systems exploiting the layer-by-layer production paradigm.…”