Abstract
NTT DATA is developing various services in the automotive field, focusing on initiatives that leverage government promotional policies to build social infrastructure through public-private cooperation. This article details NTT DATA's activities in creating a traffic-environment data portal designed to collect environmental data and match service providers with data providers. It also explores collaborative initiatives on connected-car data platforms conducted with NTT Network Technology Laboratories and NTT Communications.
The Problem: Data Silos and Technical Bottlenecks
The transition to autonomous driving and "Smart Cities" is currently hindered by data silos. Public sectors manage infrastructure (roads, signals), while private companies develop advanced services (MaaS, logistics). However, there is no unified "cooperation domain" where these parties can securely exchange data.
Furthermore, connected cars generate massive volumes of data that require low-latency processing. Standard web protocols like HTTPS are often too heavy for the high-frequency, real-time updates needed for road safety and obstacle detection. The challenge is twofold: establishing an ecosystem to match data providers with users and building a scalable, responsive technical architecture.
Methods: The MD communet and Hybrid Architecture
To address these challenges, NTT DATA employed a multi-pronged approach:
- MD communet Portal: A business-matching site that functions as a data depository. It uses semantic attributes and automatic classification to help users discover datasets through "proximity" graphs rather than just keyword searches.
- Cloud-Edge Hybrid Architecture: To handle the immediacy and volume requirements of connected cars, the platform distributes functions between central cloud servers and Multi-access Edge Computing (MEC) servers.
- Protocol Optimization: The team adopted MQTTS (Message Queueing Telemetry Transport) for vehicle-to-network (V2N) communication, as its smaller header size is significantly more efficient for high-frequency IoT and vehicle probe data than HTTPS.
- Dynamic Map Integration: High-precision 3D road maps were utilized to achieve "lane-resolution" positioning for obstacle estimation.
Key Insights: The "Cooperation Domain"
A central tenet of this research is that the mobility market cannot be won by a single player. Success requires a neutral "cooperation domain" where automakers, IT providers, and local governments can collaborate without compromising proprietary secrets.
By giving data semantic attributes, the portal can suggest datasets to users that they might not have known existed—for example, suggesting 3D road maps to a power company for utility pole maintenance. This cross-industry linkage is a cornerstone of the Society 5.0 vision.
Key Results
- Lane-Level Precision: The system successfully matched vehicle location to map information at a lane-resolution level, allowing for the detection of specific road obstacles like traffic pylons.
- Architecture Framework: The team established four criteria for evaluating connected-car use cases: Immediacy, Usage, Locality, and Data Volume.
- Functional Portal: The "MD communet" was successfully launched to facilitate both offline and online business discussions, moving beyond a simple technical repository to a social infrastructure tool.
Limitations and Future Work
Despite these successes, behavioral ambiguity remains a challenge; current algorithms struggle to differentiate intentional maneuvers from emergency ones. Future work will involve integrating multi-vehicle behavior data to improve AI accuracy. Additionally, the team aims to refine the architecture for multi-carrier support to ensure seamless data transmission across different telecommunications providers and establish mechanisms to guarantee the reliability of AI-generated safety information in real-world urban environments.
Summary updated