Enabling data exchange in two-agent interactive systems under privacy constraints

E. Veronica Belmega, Lalitha Sankar, H. Vincent Poor

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

It is advantageous for collecting agents in interconnected systems to exchange information (e.g., functions of their measurements) in order to improve their local processing (e.g., state estimation) because of the typically correlated nature of the data in such systems. However, privacy concerns may limit or prevent this exchange leading to a tradeoff between state estimation fidelity and privacy (referred to as competitive privacy). This paper focuses on a two-agent interactive setting and uses a communication protocol in which each agent is capable of sharing a compressed function of its data. The objective of this paper is to study centralized and decentralized mechanisms that can enable and sustain non-zero data exchanges among the agents. A centralized mechanism determines the data sharing policies that optimize a network-wide objective function combining the fidelities and leakages at both agents. Using common-goal games and best-response analysis, the optimal policies are derived analytically and allow a distributed implementation. In contrast, in the decentralized setting, repeated discounted games are shown to naturally enable data exchange (without any central control or economic incentives) resulting from the power to renege on a mutual data exchange agreement. For both approaches, it is shown that non-zero data exchanges can be sustained for specific fidelity ranges even when privacy is a limiting factor. This paper makes a first contribution to understanding how data exchange among distributed agents can be enabled under privacy concerns and the resulting tradeoffs in terms of leakage vs. estimation errors.

Original languageEnglish (US)
Article number7097637
Pages (from-to)1285-1297
Number of pages13
JournalIEEE Journal on Selected Topics in Signal Processing
Volume9
Issue number7
DOIs
StatePublished - Oct 1 2015

Fingerprint

Electronic data interchange
State estimation
Error analysis
Large scale systems
Network protocols
Economics
Processing

Keywords

  • Competitive privacy
  • Discounted repeated games
  • Distributed state estimation
  • Non-cooperative games

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Signal Processing

Cite this

Enabling data exchange in two-agent interactive systems under privacy constraints. / Belmega, E. Veronica; Sankar, Lalitha; Poor, H. Vincent.

In: IEEE Journal on Selected Topics in Signal Processing, Vol. 9, No. 7, 7097637, 01.10.2015, p. 1285-1297.

Research output: Contribution to journalArticle

@article{66043526736f4aa99090c37743eafb76,
title = "Enabling data exchange in two-agent interactive systems under privacy constraints",
abstract = "It is advantageous for collecting agents in interconnected systems to exchange information (e.g., functions of their measurements) in order to improve their local processing (e.g., state estimation) because of the typically correlated nature of the data in such systems. However, privacy concerns may limit or prevent this exchange leading to a tradeoff between state estimation fidelity and privacy (referred to as competitive privacy). This paper focuses on a two-agent interactive setting and uses a communication protocol in which each agent is capable of sharing a compressed function of its data. The objective of this paper is to study centralized and decentralized mechanisms that can enable and sustain non-zero data exchanges among the agents. A centralized mechanism determines the data sharing policies that optimize a network-wide objective function combining the fidelities and leakages at both agents. Using common-goal games and best-response analysis, the optimal policies are derived analytically and allow a distributed implementation. In contrast, in the decentralized setting, repeated discounted games are shown to naturally enable data exchange (without any central control or economic incentives) resulting from the power to renege on a mutual data exchange agreement. For both approaches, it is shown that non-zero data exchanges can be sustained for specific fidelity ranges even when privacy is a limiting factor. This paper makes a first contribution to understanding how data exchange among distributed agents can be enabled under privacy concerns and the resulting tradeoffs in terms of leakage vs. estimation errors.",
keywords = "Competitive privacy, Discounted repeated games, Distributed state estimation, Non-cooperative games",
author = "Belmega, {E. Veronica} and Lalitha Sankar and Poor, {H. Vincent}",
year = "2015",
month = "10",
day = "1",
doi = "10.1109/JSTSP.2015.2427775",
language = "English (US)",
volume = "9",
pages = "1285--1297",
journal = "IEEE Journal on Selected Topics in Signal Processing",
issn = "1932-4553",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "7",

}

TY - JOUR

T1 - Enabling data exchange in two-agent interactive systems under privacy constraints

AU - Belmega, E. Veronica

AU - Sankar, Lalitha

AU - Poor, H. Vincent

PY - 2015/10/1

Y1 - 2015/10/1

N2 - It is advantageous for collecting agents in interconnected systems to exchange information (e.g., functions of their measurements) in order to improve their local processing (e.g., state estimation) because of the typically correlated nature of the data in such systems. However, privacy concerns may limit or prevent this exchange leading to a tradeoff between state estimation fidelity and privacy (referred to as competitive privacy). This paper focuses on a two-agent interactive setting and uses a communication protocol in which each agent is capable of sharing a compressed function of its data. The objective of this paper is to study centralized and decentralized mechanisms that can enable and sustain non-zero data exchanges among the agents. A centralized mechanism determines the data sharing policies that optimize a network-wide objective function combining the fidelities and leakages at both agents. Using common-goal games and best-response analysis, the optimal policies are derived analytically and allow a distributed implementation. In contrast, in the decentralized setting, repeated discounted games are shown to naturally enable data exchange (without any central control or economic incentives) resulting from the power to renege on a mutual data exchange agreement. For both approaches, it is shown that non-zero data exchanges can be sustained for specific fidelity ranges even when privacy is a limiting factor. This paper makes a first contribution to understanding how data exchange among distributed agents can be enabled under privacy concerns and the resulting tradeoffs in terms of leakage vs. estimation errors.

AB - It is advantageous for collecting agents in interconnected systems to exchange information (e.g., functions of their measurements) in order to improve their local processing (e.g., state estimation) because of the typically correlated nature of the data in such systems. However, privacy concerns may limit or prevent this exchange leading to a tradeoff between state estimation fidelity and privacy (referred to as competitive privacy). This paper focuses on a two-agent interactive setting and uses a communication protocol in which each agent is capable of sharing a compressed function of its data. The objective of this paper is to study centralized and decentralized mechanisms that can enable and sustain non-zero data exchanges among the agents. A centralized mechanism determines the data sharing policies that optimize a network-wide objective function combining the fidelities and leakages at both agents. Using common-goal games and best-response analysis, the optimal policies are derived analytically and allow a distributed implementation. In contrast, in the decentralized setting, repeated discounted games are shown to naturally enable data exchange (without any central control or economic incentives) resulting from the power to renege on a mutual data exchange agreement. For both approaches, it is shown that non-zero data exchanges can be sustained for specific fidelity ranges even when privacy is a limiting factor. This paper makes a first contribution to understanding how data exchange among distributed agents can be enabled under privacy concerns and the resulting tradeoffs in terms of leakage vs. estimation errors.

KW - Competitive privacy

KW - Discounted repeated games

KW - Distributed state estimation

KW - Non-cooperative games

UR - http://www.scopus.com/inward/record.url?scp=84942251105&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84942251105&partnerID=8YFLogxK

U2 - 10.1109/JSTSP.2015.2427775

DO - 10.1109/JSTSP.2015.2427775

M3 - Article

AN - SCOPUS:84942251105

VL - 9

SP - 1285

EP - 1297

JO - IEEE Journal on Selected Topics in Signal Processing

JF - IEEE Journal on Selected Topics in Signal Processing

SN - 1932-4553

IS - 7

M1 - 7097637

ER -