[Paper Review] Increasing Data Resilience of Mobile Devices with a Collaborative Backup Service

Damien Martin-Guillerez, Michel Banâtre+1 more|arXiv (Cornell University)|Nov 3, 2006

Distributed systems and fault tolerance16 references3 citations

TL;DR

This paper proposes a transparent, collaborative backup system for mobile devices that leverages opportunistic peer-to-peer data replication among nearby devices to increase data resilience in the face of device loss or failure. By replicating data on neighboring mobile devices between global network connections, the system reduces data loss during the gap between backups, using minimal user intervention and no prior trust or specialized hardware.

ABSTRACT

Whoever has had his cell phone stolen knows how frustrating it is to be unable to get his contact list back. To avoid data loss when losing or destroying a mobile device like a PDA or a cell phone, data is usually backed-up to a fixed station. However, in the time between the last backup and the failure, important data can have been produced and then lost. To handle this issue, we propose a transparent collaborative backup system. Indeed, by saving data on other mobile devices between two connections to a global infrastructure, we can resist to such scenarios. In this paper, after a general description of such a system, we present a way to replicate data on mobile devices to attain a prerequired resilience for the backup.

Motivation & Objective

• Address the critical problem of data loss in mobile devices due to theft, loss, or failure, especially when data are created after the last backup.
• Design a backup system that operates efficiently under high mobility and intermittent connectivity, avoiding reliance on fixed infrastructure at all times.
• Enable transparent, user-transparent data backup and recovery using only existing wireless interfaces (e.g., Bluetooth, Wi-Fi) and no additional hardware.
• Ensure fair resource usage and prevent selfish behavior in peer-to-peer replication by incorporating incentives and resource-aware replication strategies.
• Support diverse mobile environments, including mobile sensor networks and collaborative robots, where data must be reliably stored and retrieved despite device mobility and failure.

Proposed method

• Use opportunistic replication: when a mobile device is disconnected from the global network, it replicates important data to nearby devices via wireless communication (e.g., Bluetooth, IEEE 802.11).
• Implement a replication ordering algorithm to prioritize which replicas are stored based on reliability, availability, and resource constraints.
• Integrate with a global backup server when connected to the Internet, allowing data to be synchronized and persisted long-term.
• Apply encryption for confidentiality and integrity, ensuring data are protected from eavesdropping, though no pre-existing trust is required between peers.
• Design a lightweight, energy-efficient protocol that avoids redundant data transfers and minimizes communication overhead during replication.
• Support data reconciliation and aggregation in systems like mobile sensor networks or collaborative robots, where local data processing reduces memory and bandwidth usage.

Experimental results

Research questions

• RQ1 How can data resilience be improved in mobile devices when the last backup was recent and the device is lost before the next synchronization?
• RQ2 What mechanisms can ensure reliable data replication among mobile peers without requiring pre-existing trust or dedicated infrastructure?
• RQ3 How can resource consumption (energy, memory, bandwidth) be minimized in a mobile, high-mobility environment during backup operations?
• RQ4 What criteria should guide the selection and deletion of replicas to maintain system efficiency and data availability?
• RQ5 How can the system be adapted to work effectively in mobile sensor networks and collaborative robotics, where devices are highly mobile and data are time-sensitive?

Key findings

• The proposed collaborative backup system significantly reduces the risk of data loss during the interval between global backups by using nearby mobile devices as temporary backup peers.
• Replication via opportunistic peer-to-peer communication is feasible and effective in high-mobility environments, especially when combined with eventual synchronization to a global server.
• The system avoids unnecessary resource consumption by not propagating replicas beyond the initial peer-to-peer transfer, reducing energy and bandwidth use.
• Replica management strategies, including deletion policies and selection criteria, are critical for maintaining system efficiency and ensuring data availability under memory and energy constraints.
• The system is compatible with existing wireless technologies (e.g., Bluetooth, Wi-Fi) and does not require additional hardware or pre-established trust relationships.
• The approach is particularly suitable for mobile sensor networks and collaborative robots, where data MULEs or mobile agents can carry and store data until they reach a base station.