Improving security in data-centric storage for wireless sensor networks

Ángel Cuevas, Azzedine BoukerchePE-WASUN 2013 · DOI

TL;DR

This paper tackles the critical security challenges in Data-Centric Storage (DCS) for Wireless Sensor Networks (WSNs), particularly when multiple applications share network resources. They propose two novel, lightweight security mechanisms: one for resource-constrained nodes that hides rendezvous node locations, and another for powerful nodes leveraging public-key cryptography, aiming to prevent unauthorized access and denial-of-service attacks without significant overhead.

What Problem Does This Paper Address?

Wireless Sensor Networks (WSNs) are increasingly used in self-managed architectures where nodes collaborate without a central authority. Data-Centric Storage (DCS) is a promising paradigm for these networks, where data is stored and retrieved based on its content rather than node location. However, when multiple applications, potentially from different entities, share the same WSN infrastructure (e.g., for routing or storage), significant security vulnerabilities emerge. Specifically, this paper addresses three key problems:

  1. Unauthorized Information Access: Malicious nodes could potentially access information belonging to applications they are not part of.
  2. Insider Threats: Nodes storing data for other applications might gain unauthorized access to that data.
  3. Denial-of-Service (DoS) Attacks: Malicious nodes could overload rendezvous nodes, disrupting the availability of data for legitimate applications.

The challenge lies in addressing these security concerns without introducing substantial overhead, which is a critical constraint for resource-limited sensor nodes.

Methods

The authors propose two distinct solutions tailored to the capabilities of sensor nodes:

  • For Resource-Constrained Sensor Nodes (Hiding Rendezvous Nodes): This approach focuses on making it harder for attackers to locate the rendezvous nodes (or home nodes in GHT). They introduce a new, application-specific parameter s into the hash function used to determine rendezvous node locations. This parameter is only known by legitimate nodes of a specific application. By requiring attackers to guess this parameter, the brute-force search for rendezvous nodes becomes significantly more difficult, thus increasing the effort and energy required for an attack. This method aims to address unauthorized access and DoS attacks by making reconnaissance challenging.
  • For Powerful Sensor Nodes (Public-Key Cryptography): For nodes with more processing power, the paper suggests leveraging Public-Key Cryptography. Each application is assigned a public-private key pair. Producer nodes encrypt data using the application's public key before sending it to rendezvous nodes. Rendezvous nodes store this encrypted data. While any node can retrieve the encrypted data, only nodes possessing the corresponding private key (i.e., legitimate application nodes) can decrypt and access the actual information. This method directly addresses unauthorized information access and insider threats. However, to prevent DoS attacks, this solution needs to be combined with a dynamic DCS approach.

The paper also emphasizes that their solutions aim to work within the existing DCS framework (put/get operations) without requiring new messages or functionalities, thus minimizing overhead.

Datasets

Not discussed in the paper. The evaluation relies on extensive simulations.

Key Insights and Ideas

  • Dynamic Rendezvous Nodes are Key for Security: The paper strongly advocates for dynamic rendezvous node selection (as in STARR-DCS) as a fundamental security feature. Changing these nodes over time significantly hinders long-term DoS attacks and makes it harder for attackers to maintain access even if they initially compromise a node.
  • Security is a Trade-off with Resource Constraints: The authors acknowledge that for highly resource-limited nodes, complex cryptographic solutions are infeasible. Their proposed method of obfuscating rendezvous node locations offers a practical, albeit not foolproof, security enhancement.
  • Layered Security Approach: The paper demonstrates the benefit of a layered security approach, where different solutions are applied based on node capabilities. Public-key cryptography provides strong confidentiality for powerful nodes, while location obfuscation offers a lighter alternative for constrained devices.
  • The Importance of Application-Specific Secrets: Introducing application-specific parameters (like s) that are not broadcast or easily discoverable is a simple yet effective way to create a barrier against unauthorized access.

Key Results

  • Resource-Constrained Solution Effectiveness: Simulations show that increasing the number of nodes in the network linearly increases the number of queries an attacker needs to find a rendezvous node. For instance, in a network of 1000 nodes, an attacker might need around 1400 queries. Similarly, increasing the number of rendezvous nodes significantly increases the queries needed for an attacker to succeed in retrieving information or launching a DoS attack.
  • DoS Attack Mitigation: The dynamic nature of STARR-DCS, where rendezvous nodes change over time, is shown to be highly effective against long-term DoS attacks. Even if an attacker compromises a node, its access is limited to the current epoch, requiring a new reconnaissance effort for subsequent epochs.
  • Public-Key Cryptography Efficacy: The public-key cryptography solution is demonstrated to effectively prevent unauthorized access to sensitive information, fulfilling requirements 1 and 2. However, it requires integration with dynamic DCS for DoS prevention.

Limitations and Future Work

  • Limitations of the Resource-Constrained Solution: The proposed solution for resource-constrained nodes does not fully guarantee that an attacker cannot access sensitive information. It primarily makes the attack more difficult and costly. It also does not address the "insider threat" (a node storing data for other applications accessing it).
  • Synchronization Assumption: The paper notes that some existing solutions (like pDCS) rely on synchronized sensors, which is a strong and often impractical assumption in WSNs. Their proposed solutions aim to avoid this.
  • Future Work: The paper suggests that for powerful nodes, the public-key cryptography solution needs to be combined with a dynamic DCS approach (like STARR-DCS) to effectively prevent long-term DoS attacks. Further research could explore more efficient key management schemes for WSNs and investigate the optimal epoch duration for dynamic rendezvous node selection.

Summary updated

BibTeX
@inproceedings{cuevas2013improving,
  author = {Ángel Cuevas and Azzedine Boukerche},
  title = {{Improving security in data-centric storage for wireless sensor networks}},
  booktitle = {PE-WASUN},
  year = {2013},
  doi = {10.1145/2507248.2507252},
}
Figure