18 lines
1.7 KiB
TeX
18 lines
1.7 KiB
TeX
\chapter{Conclusion}
|
|
The problem of leveraging power side-channel analysis to defend embedded system present unique capabilities.
|
|
In opposition with current common \gls{ids} techniques, physics-based security is not built on purpose-made actionable data.
|
|
The very nature of the input information sets this techniques aside.
|
|
Power consumption is closely related to instruction execution, and makes it a good proxy variable for machine activity.
|
|
Moreover, power is easy and cheap to measures reliably at a high sampling rate, enabling analysis of any machine consuming electricity.
|
|
Finally, a sequence of instructions is generally related to a unique power consumption pattern.
|
|
This \textit{one-to-one} relationship, allows to consider the power consumption as a signature for a software of machine activity.
|
|
However, power consumtion is not an actionable information.
|
|
From the raw time series format, little can be extracted about the machine activity or integrity.
|
|
To enable further analysis, a set of algorithm is required for both runtime online analysis as well as offline monitoring of specific activity.
|
|
The full range of capabilities is still unknown.
|
|
Successfull runtime monitoring enables the detection of activity policy violation, anomalous activity detection, machine failure detection or distributed attacks.
|
|
On the other hand, pre-OS monitoring enable the detection of boot process violation at a level where common \gls{ids} are not enabled yet.
|
|
These are just some of the possible applciation of this technology, with many more to be discovered.
|
|
|
|
With this proposal, I present some problems to study that enables some of the applications of physics-based security.
|
|
I hereby ask to continue researching these problems with the view to complete my PhD thesis.
|