PaperlayerView exact revision on arXiv

Source-linked AI summary

Privacy-preserving schemes for Ad Hoc Social Networks: A survey

Mohamed Amine Ferrag, Leandros Maglaras, Ahmed Ahmim

arXiv:1610.06095v1cs.CR

TL;DR

Existing surveys did not comprehensively cover privacy-preserving schemes for mobile and vehicular social networks. This survey examines 33 schemes and finds that surveyed work concentrates on three privacy models, while manipulation-based attacks receive less attention.

  • Problem

    Earlier surveys covered online social networks or MSN and VSN applications and architectures, but not recent privacy-preserving schemes for both network types comprehensively.

    Source: §I · p. 2

  • Method

    The survey examines 33 privacy-preserving schemes and organizes their privacy models, attack classes, countermeasures, and game-theoretic approaches.

    Source: §I · p. 3, §III · p. 6

  • Results

    70% of surveyed papers focus on location privacy, identity privacy, and anonymity; manipulation-based attacks are less considered, and only seven papers use game-theory approaches.

    Source: §VII · p. 22

  • Takeaways & Limitations

    The survey calls for evaluating VSN privacy technologies with metrics drawn from both social and vehicular domains.

    Source: C. Combination of privacy metrics · p. 23

Abstract

from arXiv · show

In this paper, we review the state of the art of privacy-preserving schemes for ad hoc social networks, including, mobile social networks (MSNs) and vehicular social networks (VSNs). Specifically, we select and in-detail examine thirty-three privacy preserving schemes developed for or applied in the context of ad hoc social networks. These schemes are published between 2008 and 2016. Based on this existing privacy preservation schemes, we survey privacy preservation models, including location privacy, identity privacy, anonymity, traceability, interest privacy, backward privacy, and content oriented privacy. The recent important attacks of leaking privacy, countermeasures, and game theoretic approaches in VSNs and MSNs are summarized in form of tables. In addition, an overview of recommendations for further research is also provided. With this survey, readers can have a more thorough understanding of research trends in privacy-preserving schemes for ad hoc social networks

I. INTRODUCTION

The survey addresses the lack of a comprehensive review of privacy-preserving schemes for mobile and vehicular social networks by examining 33 schemes and organizing their models, threats, and countermeasures. It also outlines research recommendations and reviews location-privacy approaches for both network types.

I. INTRODUCTION

  • Earlier surveys did not comprehensively cover privacy-preserving schemes for MSNs and VSNs; this survey systematically examines thirty-three schemes and presents a side-by-side comparison of the current state of the art.Source: §I · pp. 2–3

II. PRIVACY PRESERVATION MODELS

  • The reviewed privacy models include location, identity, anonymity, traceability, interest, backward, and content-oriented privacy.The survey also classifies privacy-leaking attacks and reviews countermeasures and game-theoretic approaches for MSNs and VSNs.Source: §I · p. 3

I. INTRODUCTION

  • The survey classifies privacy-leaking attacks as identity-, location-, eavesdropping-, manipulation-, or service-based.It also presents countermeasures and game-theoretic approaches for MSNs and VSNs.Source: §I · p. 3

A. Location privacy

  • For location privacy, MSN studies address proximity-based forwarding and location-sharing controls, while VSN studies use pseudonym changes, cryptography, and social-spot forwarding.The VSN approaches include conditional tracking and social-spot-based pseudonym changes; the survey describes both receiver-location privacy and packet-forwarding considerations.Source: A. Location privacy · p. 4

B. Identity privacy

The survey reviews privacy models for MSNs and VSNs and classifies privacy-leaking attacks into five categories. It covers identity privacy, anonymity, traceability, interest privacy, backward privacy, and content-oriented privacy.

B. Identity privacy

  • Identity privacy substitutes pseudonyms for real identities, while surveyed MSN and VSN schemes use approaches including anonymous credentials, group signatures, and session-specific pseudonyms.The survey also notes that cell-based communication can reduce identity-location linkage, but its per-cell servers increase cost and make that approach infeasible.Source: B. Identity privacy · p. 4

C. Anonymity- Untraceability

  • Anonymity protects user privacy and reduces impersonation risk; surveyed VSN approaches include ring signatures, onion encryption, pseudonymous authentication, and pseudonym-change analysis.The PPM protocol in MSNs supports anonymity ranging from conditional to full through comparison-based or predicate-based approaches and anonymity enhancement.Source: C. Anonymity- Untraceability · p. 5

D. Traceability

  • Traceability lets a trusted authority identify misbehaving nodes; the survey reports no current MSN study and describes VSN schemes that support tracing through identities, pseudonyms, or revocation.It suggests AMA and SAT, proposed for carpooling systems, as potentially applicable to MSNs.Source: D. Traceability · p. 5

E. Interest privacy

  • Interest privacy protects common interests used to form social-network connections, with surveyed approaches using authenticated key exchange or attribute-based encryption.FLIP addresses like-minded vehicles communicating, while Rabieh et al. use attribute-based encryption.Source: E. Interest privacy · p. 5

F. Backward privacy

  • Backward privacy requires revoked nodes to reveal no information during the revocation period; PASS uses a one-way hash function for this purpose in VSNs.The survey identifies one VSN work on backward privacy and also mentions forward and backward secrecy models.Source: F. Backward privacy · p. 5

G. Content oriented privacy

  • Content-oriented privacy is framed around immutability, transparency, and accountability, with surveyed schemes using certificate updates, certificate evolution, short signatures, and keyword-search encryption.Related models include impersonator resistance and trust evaluation.Source: G. Content oriented privacy · pp. 5–6

III. ATTACKS OF LEAKING PRIVACY

  • The survey groups privacy-leaking attacks in MSNs and VSNs into identity-based, location-based, eavesdropping-based, manipulation-based, and service-based categories.It uses this five-part classification alongside summaries of threats in the two network types.Source: §III · p. 6

A. Identity-based attack

  • Identity-based attacks manipulate legitimate users’ identities: Sybil attacks use multiple identities, while impersonation records a vehicle’s pseudonym during registration.The survey describes countermeasures including multiple reviews or valid pseudo-IDs for Sybil detection, and message authentication or tracking and revocation for impersonation.Source: A. Identity-based attack · pp. 6–7

B. Location-based attack

The survey distinguishes location-based attacks from attacks that exploit eavesdropping on network communications. It describes attack mechanisms and summarizes approaches proposed to detect or resist several of them.

B. Location-based attack

  • Location-based attacks seek to reveal user locations and include forgery and global external attacks.Forgery uses bogus location information to initiate location tracking; a global external adversary tracks vehicles by time, location, and velocity.Source: B. Location-based attack · pp. 7–8
  • The survey describes countermeasures for forgery and global external attacks, including profile comparison, authentication, group signatures, and pseudonym generation with conditional tracking.The PPM protocol is described as detecting profile forgery, while the PCS strategy uses pseudonym self-delegated generation and conditional tracking.Source: B. Location-based attack · p. 8

C. Eavesdropping-based attack

  • Eavesdropping-based attacks target network communications and include eavesdropping, packet analysis, packet tracing, and hole attacks.A hole attack creates a communication tunnel through which adversaries eavesdrop; multiple adversaries may control routing through that tunnel.Source: C. Eavesdropping-based attack · pp. 8–9
  • Eavesdropping and packet analysis can expose transmitted data or common interests; the survey cites delegated encryption, semantic security, authorization, and anonymous authentication as defenses.The SPRING protocol uses conditional privacy-preserving authentication with privacy-preserving authentication and conditional tracking phases against packet analysis.Source: C. Eavesdropping-based attack · p. 8
  • Packet tracing reveals packet routes without recovering content, while attribute tracing can expose information; the survey describes SPRING’s anonymous authentication as a defense.The survey connects packet tracing to routing and demographic-inference attacks.Source: C. Eavesdropping-based attack · p. 9
  • Hole-attack countermeasures include cooperative-neighbor techniques, proxy re-signature, anonymous identity-based encryption, and detection or tracking methods.EPSA’s wormhole detection rate depends on tunnel length, with longer tunnels associated with higher detection rates.Source: C. Eavesdropping-based attack · pp. 9–10, Table VI · p. 9

D. Manipulation-based attack

The survey covers manipulation-based and service-based attacks, then reviews cryptographic countermeasures and game-theoretic approaches. It notes that most examined privacy-preserving schemes use cryptography.

D. Manipulation-based attack

  • Manipulation-based attacks target users or socialspots through user manipulation or man-in-the-middle attacks.User manipulation can make a node treat an adversary as a hotspot; a cited protocol uses authentication and an anonymous privacy-preserving routing tree, while EP2DF addresses man-in-the-middle attacks with a lite-CA-based public-key cryptosystem.Source: D. Manipulation-based attack · p. 10

E. Service-based attack

  • Service-based attacks aim to make network services unavailable and include replay, denial-of-service, source-bogus, and spam attacks.The survey describes timestamp, sequence-number, and pseudo-identity checks against replay; PASS uses Schnorr signatures and prestored certificates against DoS, STAP addresses source-bogus attacks, and LENS uses gatekeepers against spam.Source: E. Service-based attack · pp. 10–11

IV. COUNTERMEASURES AND GAME THEORETIC APPROACHES

  • Most examined privacy-preserving schemes use cryptography as a countermeasure, and the survey discusses game-theoretic models for preserving privacy in MSNs and VSNs.It points to tables summarizing cryptographic methods used in the two network types.Source: §IV · p. 11, Table VIII · p. 11, Table IX · p. 12

A. Symmetric-key primitives

  • Symmetric-key primitives comprise symmetric-key ciphers, keyed hash functions, and pseudorandom sequences; HMAC is used for message integrity and source-data authentication.The cited scheme assigns 128-bit secret keys to users and the authority for its HMAC computation.Source: A. Symmetric-key primitives · p. 11

B. Public-key primitives

Public-key primitives are widely used in the surveyed privacy-preserving schemes. The section describes encryption and signature techniques for protecting identities, messages, and access to information.

B. Public-key primitives

  • Public-key primitives are used by most privacy-preserving schemes covered in the survey.The section groups them into public-key encryption and signatures.Source: B. Public-key primitives · p. 12
  • Schemes,, and use identity-based encryption; upon OBU registration with identifier IDi, the proposed scheme uses secret key s to encrypt IDi into pseudo-ID PIDi = Encss(IDi||ri), where ri is randomly chosen from Z∗.Source: B. Public-key primitives · p. 12
  • Other encryption approaches include keyword-search, attribute-based, and hidden-vector encryption for document search, attribute-based access, and fine-grained filtering,,.The hidden-vector scheme represents a fine-grained keyword filter as a vector.Source: B. Public-key primitives · p. 12
  • Certificate-less public-key encryption and proxy re-encryption provide key-management and ciphertext re-encryption operations,.The certificate-less scheme includes key-generation-center algorithms; proxy re-encryption creates a re-encryption key for another identity.Source: B. Public-key primitives · pp. 12–13
  • Homomorphic encryption supports an additive operation on ciphertexts, so multiplying encryptions yields an encryption of the summed plaintexts,,.The survey also describes signatures used for anonymous authentication, including short signatures and group signatures,,,,,,,.Source: B. Public-key primitives · p. 13

C. Unkeyed primitives

The section describes cryptographic hashes for message integrity and reviews a small set of game-theoretic approaches applied to privacy, forwarding, authentication, and attack defense in MSNs and VSNs.

C. Unkeyed primitives

  • Secure cryptographic hash functions are used in most surveyed schemes to check message integrity and are characterized by pre-image, second pre-image, and collision resistance.The section defines the hash output length using security parameter λ.Source: C. Unkeyed primitives · pp. 13–14

D. Game theoretic approaches

  • Only a small fraction of surveyed schemes use game theory, with approaches including static, cooperation, sequence, non-cooperative, and traceability games.Applications include location privacy, data forwarding, authentication, protocol security, and group-signature traceability.Source: D. Game theoretic approaches · p. 14
  • Game-theoretic approaches also target wormhole detection and prevention, while reported caveats include unmodeled diverse behavior and absent comparisons with other methods.Other listed limitations include no comparison with available spam-prevention schemes and many assumptions needed to understand implementation.Source: D. Game theoretic approaches · p. 15
  • Game-theoretic MSN schemes address privacy-preserving cooperative forwarding, shared health information, and secure same-symptom handshaking.The handshaking scheme is also reported to consider delivery ratio and reporting delay, while lacking a thorough comparison with similar methods.Source: D. Game theoretic approaches · p. 15
  • Other schemes cover trust evaluation and spam prevention, privacy controls, fine-grained filtering, and emergency-service availability with transaction unlinkability.The fine-grained filtering approach includes social-assisted distribution and Merkle hash tree-based filter authentication and updates.Source: D. Game theoretic approaches · p. 15
  • Some schemes report performance or comparison measures but retain scope caveats, including no comparison with other frameworks and no consideration of identity or location privacy.The survey also notes a need for large Detectreq reporting and that mobility models are not considered in one scheme.Source: D. Game theoretic approaches · pp. 14–15

V. PRIVACY-PRESERVING SCHEMES FOR MSNS

The section examines fourteen privacy-preserving schemes for MSNs, classifies them into six categories, and notes that they were published from 2011 to 2016.

V. PRIVACY-PRESERVING SCHEMES FOR MSNS

  • Fourteen MSN schemes published between 2011 and 2016 are classified as social profile, social morality, social routing, social health, location-based services, or service-oriented sociality.Source: §V · p. 16

A. Network model with social profile

The surveyed schemes address social-profile matching and morality-aware forwarding as distinct privacy-preserving problems in mobile social networks.

A. Network model with social profile

  • PPM lets users compare profiles without disclosing them, while supporting non-anonymity, conditional anonymity, and full anonymity.Its comparison uses explicit comparison, implicit comparison, and implicit predicate phases; the survey notes limited analysis of how “=” affects anonymity.Source: A. Network model with social profile · p. 16

B. Network model with social morality

  • A morality-driven forwarding protocol combines route-based authentication and proximity measurement to protect user location and identity privacy.The survey notes that diverse behavior models are not considered.Source: B. Network model with social morality · p. 16

C. Network model with social routing

Across the surveyed social-network models, privacy schemes target routing, health data, location services, and service-oriented interaction, with varied mechanisms and stated limitations.

C. Network model with social routing

  • ECPDR, SDPP, and EPSA add privacy or attack defenses to social routing, but the surveyed schemes carry limits in adversary analysis, report size, or network assumptions [40],,.ECPDR targets black-hole attacks, SDPP provides transparency and impersonator resistance, and EPSA detects and avoids wormhole attacks.Source: C. Network model with social routing · p. 16

D. Network model with social health

  • HealthShare, PEC, and SSH address privacy or secure interaction in health social networks, while their evaluations note missing comparisons or a need for decentralized emergency response,,.PEC supports transaction unlinkability; SSH uses a same-symptom-based handshake.Source: D. Network model with social health · pp. 16–17

E. Network model with location-based services

  • SmartMask addresses location privacy against demographic inference using common-trace and machine-learning approaches, but the survey says it needs comparison with other frameworks.The survey also notes that the social-spot idea in VSLP could apply to this category.Source: E. Network model with location-based… · p. 17

F. Network model with service-oriented sociality

  • Service-oriented schemes include trust evaluation against Sybil attacks and fine-grained filtering, with the PIF scheme’s filter updates described as efficient,,.The survey notes missing system comparisons for TSE and strong network-characteristic assumptions for PIF.Source: F. Network model with service-oriented… · p. 17
  • VANET schemes in the tabulated survey cover authentication, anonymity, conditional privacy, backward privacy, and identity revocation.The table entries also identify location privacy as unaddressed in some schemes and mobility models as unconsidered in others.Source: F. Network model with service-oriented… · p. 19
  • Interest-focused VANET protocols support like-minded-vehicle communication while aiming to protect interest privacy from vehicles without the same interest.The table describes FLIP-style chatting and interest verification, alongside a reported average delay for finding a like-minded vehicle.Source: F. Network model with service-oriented… · pp. 19–20
  • The tabulated evaluations report delivery, delay, and overhead measures, while also flagging limited threat-model analysis, absent protocol comparisons, and omitted location privacy.These limitations appear alongside reported data-transmission and invalid-batch-success-rate measures.Source: F. Network model with service-oriented… · pp. 18–20

VI. PRIVACY-PRESERVING SCHEMES FOR VSNS

The survey examines nineteen privacy-preserving schemes developed for or applied to vehicular social networks, classifying them by network model.

VI. PRIVACY-PRESERVING SCHEMES FOR VSNS

  • Nineteen VSN privacy-preserving schemes published between 2008 and 2016 are examined and classified into social spots, location-based services, and service-oriented sociality.Source: §VI · p. 20

A. Network model with social spots

The survey organizes social-spot models and reviews VSN schemes that use them for location privacy, pseudonym changes, and vehicle communication.

A. Network model with social spots

  • The survey distinguishes three social-spot models: spots integrated with RSUs, small and large spots, and global and individual spots.Source: A. Network model with social spots · p. 20
  • PCS uses key generation, self-delegated pseudonym generation, and conditional tracking to pursue location privacy, with feasibility assessed through game-theoretic techniques.The survey notes limited analysis of different threat models.Source: A. Network model with social spots · p. 20
  • STAP uses social-tier assistance, while SPF uses a “Sacrificing the Plum Tree for the Peach Tree” tactic for location privacy,.The survey reports packet-delivery and delay efficiency for STAP, and average packet-delivery ratio and delay efficiency for SPF; SPF lacks comparison with similar protocols.Source: A. Network model with social spots · pp. 20–21
  • MixGroup uses global and individual social spots to support pseudonym changes and improve users’ location privacy.Source: A. Network model with social spots · p. 21

B. Network model with location-based services

The survey reviews privacy-preserving schemes for location-based services and service-oriented sociality in VANETs, describing their security goals and reported efficiency measures.

B. Network model with location-based services

  • DIKE uses cooperative V-2-V key updates to provide forward and backward secrecy and collusion resistance, with key-update delay and ratio as efficiency measures.For parking services, Lu et al. describe a privacy-preserving intelligent parking method, while Xiong et al. propose multi-level conditional privacy with storage and computational efficiency.Source: B. Network model with location-based… · p. 21
  • EP2DF combines lite-CA-based and identity-based public-key cryptosystems, while PASS is reported efficient in revocation, certificate updating, and authentication overhead.The passage compares PASS with ECPP and DCS but does not provide numerical results.Source: B. Network model with location-based… · p. 21

C. Network model with service-oriented sociality

  • SPRING uses an intersection vertex’s Social Degree to optimize RSU deployment and vehicular DTN, whereas FLIP supports shared-interest communication while protecting interest privacy from non-matching vehicles.SPRING is reported efficient in delivery ratio and packet delay; FLIP in delay for finding like-minded vehicles.Source: C. Network model with service-oriented… · p. 21
  • VSPN provides distributed destination guidance with driver privacy, while SECSPP addresses secure V-2-V and vehicle-to-roadside communication scenarios.VSPN is reported efficient in processing delay and travel-time reduction; SECSPP in computational, communication, and storage overhead.Source: C. Network model with service-oriented… · p. 21
  • PACP targets authentication and privacy through low-latency pseudonym generation, scalability, and revocation; SPECS and PPBMA are assessed on transmission and delay measures.SPECS is described in terms of data transmission and invalid-batch success rate; PPBMA addresses link-layer privacy-preserving broadcast.Source: C. Network model with service-oriented… · p. 21

VII. RECOMMENDATIONS FOR FURTHER RESEARCH

The survey identifies imbalances and open issues in privacy-preserving research for ad hoc social networks, then recommends directions spanning methods, metrics, vulnerabilities, and human factors. Its conclusion emphasizes that users and networks are interdependent and that human factors remain relevant to protection.

VII. RECOMMENDATIONS FOR FURTHER RESEARCH

  • 70% of surveyed papers focus on location privacy, identity privacy, and anonymity; 95% use both public-key and unkeyed algorithms, while manipulation-based attacks receive less attention.The survey reports an average publication date of 2012 and only seven papers using four types of game-theory approach.Source: §VII · p. 22

A. Privacy-preserving methods

  • The proposed future directions include privacy-preserving energy consumption, privacy for V2G social networks, and privacy for the social internet of vehicles.The survey frames energy management under social clustering as an open question and notes that the cited V2G survey does not study social characteristics.Source: A. Privacy-preserving methods · pp. 22–23

B. Interdepedent privacy

  • Interdependent privacy concerns how one user’s privacy is affected by others’ decisions, a salient issue in social networks and VSNs with unknown interacting entities.The passage gives third-party applications and vehicular-social applications combined with traditional online social networks as examples of potential privacy threats.Source: B. Interdepedent privacy · p. 23

C. Combination of privacy metrics

  • Because VSNs combine social and vehicular networks, evaluations of privacy-enhancing technologies should select metrics from both domains.The passage calls for thorough evaluation of new technologies and combinations of existing technologies.Source: C. Combination of privacy metrics · p. 23

D. Identification of areas of vulnerability

  • The survey calls for identifying critical and vulnerable nodes in heterogeneous networks, while also accounting for human factors and evaluating privacy metrics in real environments.It notes that human error underlies half of the cited breaches and that real deployments are limited, especially for VSNs; it also identifies quantum privacy approaches as a research direction.Source: D. Identification of areas of… · p. 23

VIII. CONCLUSION

  • The conclusion synthesizes 33 schemes, privacy models, threats, countermeasures, and game-theoretic models, and argues that interdependencies, vulnerability identification, combined metrics, and human factors matter to protection.It states that privacy preservation in MSNs and VSNs remains challenging because adversaries can exploit system vulnerabilities.Source: §VIII · p. 23
Loading 1610.06095v1…