Introduction
In the early 2000s, the vision of "ubiquitous computing"—where sensors and devices seamlessly integrate into our daily lives—began transitioning from theoretical research to practical implementation. While initial research focused heavily on the functionality of individual context-aware services (CAS), such as a smartphone that silences itself in a theater, a significant gap emerged regarding the operational management of these systems. This paper argues that the primary hurdle for ubiquitous computing is not the technology itself, but the complex task of managing highly dynamic, multi-provider services across organizational boundaries.
The Management Gap in Ubiquitous Computing
In a real-world ubiquitous environment, context data (e.g., location, heart rate, or weather) is sourced from dozens of different providers, changes instantly due to mobility, and must be processed in real-time. Traditional, hierarchical management systems are insufficient because no single entity owns the entire infrastructure. The paper addresses how to manage these services while maintaining security, reliability, and fair billing in a decentralized ecosystem.
To illustrate these challenges, the authors utilize a hypothetical Medical Advice and Emergency System (MAES). This system must integrate vital sign sensors, medical profiles, weather databases, and traffic sensors to provide real-time advice to patients and coordinate with emergency services.
The Context Value Chain
The authors introduce the Context Value Chain, a process-oriented model that structures context procurement into four distinct stages. This model helps identify where management intervention is required to ensure data quality and service continuity.
- Sensing: The capture of raw data from physical or virtual sensors.
- Refinement: The transformation of raw data into high-level information (e.g., converting GPS coordinates into a specific street address).
- Aggregation: The combination of various refined data points relevant to a specific entity, such as a patient.
- Contextualization: The application of aggregated data to adapt the service's behavior, such as triggering an emergency alert.
Federated Management and Role Models
Because CAS involves multiple independent actors—including network operators, sensor owners, and service providers—management cannot be hierarchical. The authors propose a Federative Role Model to define how these entities interact and negotiate without a central authority.
Key roles include:
- Context Provider: Supplies refined context data.
- Context Broker: Acts as an intermediary to discover and negotiate context sources.
- Context Owner: The subject of the context (usually the user) who must maintain control over their data.
- CAS Provider: The entity delivering the final service to the user.
Key Management Innovations
The paper highlights several conceptual shifts required for effective CAS management:
- The Micro-Lifecycle: Unlike traditional services with long lifecycles, CAS requires a "micro-lifecycle" where service instances are created, configured, and torn down ad-hoc, potentially lasting only as long as a user remains in a specific location.
- Quality of Context (QoC): Similar to Quality of Service (QoS) in networking, QoC measures parameters like accuracy, timeliness, and validity. Management systems must use QoC to choose between competing data sources (e.g., choosing between precise GPS and cheaper cell-tower triangulation).
- Price-Driven Provisioning: The authors suggest a radical shift where the user's "price ceiling" might dynamically define the service quality, forcing the management system to build the best possible context value chain within a specific budget.
Applying the FCAPS Model
The paper maps CAS management challenges onto the traditional FCAPS (Fault, Configuration, Accounting, Performance, Security) framework:
- Fault Management: Systems must distinguish between a sensor failure and a "temporary absence" (e.g., losing GPS signal in a tunnel) to trigger alternative data sourcing.
- Configuration: Focuses on the ad-hoc creation of service instances based on shifting user situations.
- Accounting: Requires inter-provider accounting to support "service roaming," allowing users to move between different context providers seamlessly.
- Performance: Emphasizes context caching and reuse to meet strict real-time requirements.
- Security: Prioritizes "informational self-determination," ensuring users can authorize or prohibit access to their data even when generated by third-party sensors.
Conclusion and Future Directions
While providing a robust conceptual roadmap, the authors acknowledge that significant work remains. Future research must focus on developing a standardized Context Description Language (CDL) for automated interoperability, formalizing calculable trust models for service negotiation, and advancing self-management concepts to achieve zero-human-interaction procurement loops.
Summary updated