TL;DR
This paper introduces a novel "user-centric" incentive framework for participatory mobile phone sensing. Instead of relying on server-driven payments that often favor the server, it empowers users to control their participation level (LoPS) and dynamically adjust sensing costs, thereby maximizing their own benefits and encouraging participation.
What Problem Does This Paper Address?
Mobile phones are packed with sensors, offering immense potential for collective sensing applications in areas like environmental monitoring, healthcare, and social interaction. However, getting individual users to consistently contribute their phone's resources (battery, storage, privacy) is a major hurdle. Current incentive models are often "application-centric," meaning they're designed to benefit the sensing server by minimizing payments to users. This approach ignores the user's runtime flexibility and diverse cost tolerances, leading to unrealistic participation assumptions and potential user disengagement. The paper aims to bridge this gap by creating a system that prioritizes user benefits and flexibility.
Methods
The core of the proposed solution is a user-centric incentive framework deployed as middleware on the smartphone. This framework includes:
- Universal Sensing Platform: This platform acts as an intermediary between the operating system and sensing applications. It takes sensing task requirements and a user's chosen "Level of Participation in Sensing" (LoPS) as input.
- Task Quantifier: Translates heterogeneous sensing task requirements into a generic, dimensionless metric representing sensing performance quality.
- Resource Monitor: Continuously tracks the smartphone's available resources (battery, storage) to assess sensing costs.
- Sensing Planner: Generates a task-independent "sensing plan" that balances sensing costs and performance based on the user's LoPS and current resource conditions. This plan dictates reduced sensing performance requirements to minimize user costs.
- Incentive Design: The framework then leverages the generated sensing plan to reduce three main categories of sensing costs, thereby incentivizing users:
- Energy Saving: Exploits temporal correlations in sensor readings to reduce sensing frequency without significantly impacting performance, based on the sensing plan.
- Local Storage Conversation: Implements cooperative caching techniques that prioritize caching popular data items at strategically chosen locations (Network Central Locations - NCLs) based on data popularity and user LoPS.
- Privacy Preservation: (Mentioned as a category of sensing costs, but details are not elaborated in the provided text).
Datasets
Not discussed in the paper.
Key Insights and Ideas
- User-Centricity is Key: Shifting from server-centric payments to user-centric control over participation levels (LoPS) is crucial for sustained user engagement.
- LoPS as a Flexible Lever: Allowing users to dynamically adjust their LoPS provides runtime flexibility, enabling them to balance their needs (battery, storage, privacy) with the demands of sensing tasks.
- Cost Reduction as Incentive: Instead of direct payments, reducing sensing costs (energy, storage) for the user acts as a direct incentive, aligning with the rational self-interest of users.
- Generic Sensing Plan: The ability to translate diverse sensing tasks into a universal, task-independent sensing plan is a significant step towards a generalized sensing framework.
- Exploiting Data Redundancy: Temporal correlations in sensor data offer a rich opportunity to reduce energy consumption by intelligently omitting redundant readings.
Key Results
The paper outlines the framework and its components, demonstrating how it can reduce sensing costs. Specific quantitative results are not presented in this excerpt, as it focuses on the design and conceptualization. However, the examples illustrate the potential:
- User A, with a fully charged phone, can maintain high-quality sensing (within 30% energy consumption).
- User B, with limited storage and privacy concerns, can opt for a low LoPS, reducing sensing costs to below 20% and ensuring normal smartphone use.
Limitations and Future Work
- Probabilistic Sensing Planning: The authors acknowledge that their current probabilistic framework for sensing planning might not be sufficient for dynamic resource condition changes. Future work will involve devising optimization techniques for adaptive sensing plan adjustments.
- Dynamic Adaptation: The sensing plan needs to be more adaptive to rapid changes in system resource conditions to prevent resource depletion.
- Privacy Preservation Details: While privacy is listed as a cost category, the specific mechanisms for privacy preservation are not detailed in this excerpt.
- Extensive Experiments: The authors plan to conduct extensive experiments to evaluate the effectiveness of their user incentive design in realistic mobile sensing applications with actual users.
- Substantiating LoPS Interpretation: Future work will further substantiate the design of the sensing planner by addressing research issues related to the interpretation of LoPS.
Summary updated