The use of programmable data planes in network research has led to great innovation in the last decade. With programmable data planes, a network administrator can customise and monitor the lowest-level behaviour of network devices via standard open-specific APIs. This offers great flexibility and control to the people operating the network. Research demonstrates that many applications can be transferred from servers to devices such as switches or network cards, taking advantage of the data plane programmability offered by these devices. Along that line of research on programmable data planes, this work proposes the implementation of a privacy-preserving mechanism in the network data plane, leveraging the flexibility of programmable switches and the expressiveness of the domain-specific programming language P4. We choose Differential Privacy (DP) as the target privacy-preserving technique, hence, orienting our data plane design to DP-based mechanisms. Mainly, we show how to implement and assemble Floating Point (FP) operations on P4 targets to build a differential privacy mechanism working on a vector of input elements. Even though data plane programmability is a very active research area, at present, we are not aware of any other work in the literature that presents the implementation of in-switch differential privacy technique. We believe that our initial effort sheds some important light on the challenges and trade-offs to devise such in-network functionality.

The use of programmable data planes in network research has led to great innovation in the last decade. With programmable data planes, a network administrator can customise and monitor the lowest-level behaviour of network devices via standard open-specific APIs. This offers great flexibility and control to the people operating the network. Research demonstrates that many applications can be transferred from servers to devices such as switches or network cards, taking advantage of the data plane programmability offered by these devices. Along that line of research on programmable data planes, this work proposes the implementation of a privacy-preserving mechanism in the network data plane, leveraging the flexibility of programmable switches and the expressiveness of the domain-specific programming language P4. We choose Differential Privacy (DP) as the target privacy-preserving technique, hence, orienting our data plane design to DP-based mechanisms. Mainly, we show how to implement and assemble Floating Point (FP) operations on P4 targets to build a differential privacy mechanism working on a vector of input elements. Even though data plane programmability is a very active research area, at present, we are not aware of any other work in the literature that presents the implementation of in-switch differential privacy technique. We believe that our initial effort sheds some important light on the challenges and trade-offs to devise such in-network functionality.

Designing a privacy preserving solution on programmable switches

COREGGIOLI, CRISTIAN
2023/2024

Abstract

The use of programmable data planes in network research has led to great innovation in the last decade. With programmable data planes, a network administrator can customise and monitor the lowest-level behaviour of network devices via standard open-specific APIs. This offers great flexibility and control to the people operating the network. Research demonstrates that many applications can be transferred from servers to devices such as switches or network cards, taking advantage of the data plane programmability offered by these devices. Along that line of research on programmable data planes, this work proposes the implementation of a privacy-preserving mechanism in the network data plane, leveraging the flexibility of programmable switches and the expressiveness of the domain-specific programming language P4. We choose Differential Privacy (DP) as the target privacy-preserving technique, hence, orienting our data plane design to DP-based mechanisms. Mainly, we show how to implement and assemble Floating Point (FP) operations on P4 targets to build a differential privacy mechanism working on a vector of input elements. Even though data plane programmability is a very active research area, at present, we are not aware of any other work in the literature that presents the implementation of in-switch differential privacy technique. We believe that our initial effort sheds some important light on the challenges and trade-offs to devise such in-network functionality.
2023
Designing a privacy preserving solution on programmable switches
The use of programmable data planes in network research has led to great innovation in the last decade. With programmable data planes, a network administrator can customise and monitor the lowest-level behaviour of network devices via standard open-specific APIs. This offers great flexibility and control to the people operating the network. Research demonstrates that many applications can be transferred from servers to devices such as switches or network cards, taking advantage of the data plane programmability offered by these devices. Along that line of research on programmable data planes, this work proposes the implementation of a privacy-preserving mechanism in the network data plane, leveraging the flexibility of programmable switches and the expressiveness of the domain-specific programming language P4. We choose Differential Privacy (DP) as the target privacy-preserving technique, hence, orienting our data plane design to DP-based mechanisms. Mainly, we show how to implement and assemble Floating Point (FP) operations on P4 targets to build a differential privacy mechanism working on a vector of input elements. Even though data plane programmability is a very active research area, at present, we are not aware of any other work in the literature that presents the implementation of in-switch differential privacy technique. We believe that our initial effort sheds some important light on the challenges and trade-offs to devise such in-network functionality.
P4
Network security
Programmable Switch
File in questo prodotto:
File Dimensione Formato  
Coreggioli_Cristian.pdf

accesso aperto

Dimensione 2.23 MB
Formato Adobe PDF
2.23 MB Adobe PDF Visualizza/Apri

The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12608/64780