Detection of hardware based attacks using performance counters

Hardware attacks, such as Spectre or Meltdown, try to exploit vulnerabilities on the processor architecture, usually to gain access to privileged memory locations, producing leaks of critical information. Most of this attacks exploit out-of-order execution and also highly stress the computer’s cache...

Descripción completa

Detalles Bibliográficos
Autores: Andreu Gerique, David, Alonso García, Martí, Costa Prats, Juan José|||0000-0003-2479-0230, Morancho Llena, Enrique|||0000-0003-2403-8145, Canal Corretger, Ramon|||0000-0003-4542-204X, Otero Calviño, Beatriz|||0000-0002-9194-559X
Tipo de recurso: informe técnico
Fecha de publicación:2024
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/408859
Acceso en línea:https://hdl.handle.net/2117/408859
Access Level:acceso abierto
Palabra clave:Computer crimes
Support vector machines
Delictes informàtics
Àrees temàtiques de la UPC::Informàtica::Seguretat informàtica
Descripción
Sumario:Hardware attacks, such as Spectre or Meltdown, try to exploit vulnerabilities on the processor architecture, usually to gain access to privileged memory locations, producing leaks of critical information. Most of this attacks exploit out-of-order execution and also highly stress the computer’s cache, and both of these behaviours can be tracked using hardware performance counters (HPC). In order to try to use HPCs in our advantage we: 1. Collected several samples of different attacks and HPCs 2. Collected an equivalent amount of samples from benign programs 3. Trained a Support Vector Machine to classify the execution as benign or malignant