**6. Conclusions**

Novel non-invasive sensors were developed to collect data for analysis of analog, mixed-signal, power, and EM signal behavior. To sense small changes in magnetic fields and inform the machine learning circuits, the nanoscale heterostructure was developed to be able to monolithically integrate CMOS circuits with novel spin-torque devices that can be utilized as robust high-fidelity sensors and embedded into interconnects. Lightweight learning algorithms were developed for fast threat detection at the front-end of resourceconstraint devices in real time. The MTJ sessors were fabricated, measured and modeled for Cadence simulations together with the presented attacker models. The results show that the proposed system achieves 95% of the accuracy to recognize the attacker with all trojan types applied.

**Author Contributions:** Conceptualization, E.C., J.Z. and V.C.; methodology, E.C., J.K. and B.-Y.Y.; software, J.K.; validation, J.K. and B.-Y.Y.; formal analysis, E.C., J.K. and B.-Y.Y.; investigation, E.C., J.K. and B.-Y.Y.; resources, J.Z. and V.C.; data curation, E.C. and J.K.; writing—original draft preparation, E.C., J.K. and B.-Y.Y.; writing—review and editing, J.Z. and V.C.; visualization, E.C., J.K. and B.-Y.Y.; supervision, J.Z. and V.C.; project administration, J.Z. and V.C.; funding acquisition, J.Z. and V.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the National Science Foundation under Grant No. 1952907, 1953801, and 2028893, and the Data Storage Systems Center (DSSC) at Carnegie Mellon University.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** This study did not report any data.

**Conflicts of Interest:** The authors declare no conflict of interest.
