Introduction
Social Sensing based Edge Computing (SSEC) represents a paradigm shift in how we handle the massive influx of data from our physical environment. Traditionally, social sensing—the collection of data via humans or their mobile devices—has relied on centralized cloud backends for analysis. However, this model faces critical challenges: high latency, bandwidth bottlenecks, and the underutilization of powerful hardware in modern consumer electronics.
SSEC proposes "marrying" social sensing with edge computing to create a decentralized, human-centric architecture. By shifting processing to the privately-owned devices (smartphones, IoT, drones) that collect the data, SSEC leverages untapped compute cycles while reducing the burden on core networks.
The SSEC Architecture
The authors define the SSEC framework across three distinct layers designed to bridge the gap between the physical world and global services:
- Edge Layer: Comprised of privately owned "social sensors" (e.g., smartphones, wearables, drones) that perform initial sensing, storage, and local computation.
- Edge Server Layer: Intermediate nodes such as cloudlets or smart routers that act as a bridge, providing local data aggregation and service management.
- Service Layer: The backend cloud that provides a global interface, performing high-level data analytics and task release.
System Variations
To accommodate diverse real-world requirements, the paper outlines three primary system variations for SSEC deployment:
- Hierarchical: A structured flow where devices offload data to local edge servers, which then report to the cloud.
- Collaborative Edge: Devices in close proximity self-organize into peer-to-peer clusters to share workloads without requiring a dedicated server.
- Hybrid: A combination of both models, particularly useful in disaster response scenarios where local infrastructure may be unreliable.
Key Insights: The "Rational Edge"
A core contribution of this vision is the concept of the Rational Edge. Unlike traditional distributed systems where nodes are assumed to be cooperative, SSEC nodes are owned by individuals with competing objectives. A user may prioritize battery life over task speed, leading to the "selfish node" problem. Consequently, SSEC must incorporate robust incentive mechanisms and account for asymmetric information—where the server may not know a device's current battery status or privacy constraints.
Furthermore, SSEC envisions humans as more than just data sources. In complex scenarios like identifying abnormal events in a crowd, human intelligence can act as a high-accuracy cognitive processing node within the edge network, creating a true human-machine hybrid system.
Advantages and Future Directions
By processing data locally, SSEC offers several key advantages:
- Privacy through Locality: Sensitive data (like faces or license plates) can be processed on-device, with only anonymized features or alerts sent to the cloud.
- Unprecedented Coverage: Leveraging billions of existing private devices worldwide.
- Real-time Responsiveness: Significant delay reduction by processing data at the source.
Despite its potential, several challenges remain. There is a pressing need for middleware that can abstract the extreme hardware heterogeneity of billions of devices. Additionally, developing formal game-theoretic models for incentive design and integrating human cognitive skills into digital pipelines remain open research questions. As 5G matures, the high density of base stations will provide new opportunities and management challenges for the SSEC paradigm.
Summary updated