Industrial IoT applications depend on sensor data being both available and credible. Yet wireless sensor nodes may operate in unattended or harsh environments, where compromised or unreliable devices can send misleading readings. Trustworthy data collection offers a way to address this risk while accounting for the limited energy and computing capacity of many sensor nodes. An edge-based model combines node assessment with mobile collection-path planning, helping the collector prioritize more trusted sources rather than treating every sensor as equally reliable.
Why trustworthy data collection matters at the edge
Sensor networks provide the measurements used by higher-level IoT systems and applications. If those measurements are invalid, downstream processing and decisions may also be compromised. External security measures, such as authentication and secure routing, can help protect network communications, but they may not adequately address malicious behavior originating within the network. Trust evaluation provides a complementary approach: it uses observed behavior and node characteristics to estimate whether a sensor should be relied upon.
Collection itself also has operational constraints. Conventional approaches may require a mobile collector to visit many underlying nodes, including those whose data is untrustworthy. More visits can increase collection time and put additional demands on energy-limited sensors. A collector built from a resource-constrained sensor node may also lack the computing and storage capacity needed for more demanding collection tasks.
Edge computing changes where some of this work can be handled. In the described model, a mobile edge node acts as the data collector. Its role is to gather data near the sensor network, rather than requiring ordinary sensor nodes to take on the collector’s stronger computing and storage responsibilities.
How sensor node trust evaluation works
The model assigns each sensor a quantified trust value on a scale from 0 to 10. A value of 5 represents the initial, neutral state; values above 5 indicate a trustworthy node, while values below 5 indicate an untrustworthy one. Using an integer interval is intended to reduce the memory needed to represent and communicate trust values compared with storing real-number values.
Trust is assessed from several dimensions rather than a single indicator. For directly observed relationships, the assessment considers:
- Communication behavior: the history of successful communication between nodes.
- Positional intimacy: the distance between a sensor and its cluster head, which affects the likelihood of successful communication.
- Energy surplus: the node’s remaining energy relative to its initial energy.
- Packet delivery behavior: packet loss, used as an indicator of link or transmission behavior.
These inputs are combined using weights, allowing the model to account for multiple aspects of node behavior and condition. The intent is to avoid basing trust on a narrow measure that could give an incomplete picture.
Using indirect trust when direct observation is limited
A node may not have enough direct interaction with another sensor to assess it confidently. In that case, the model can use recommendations relayed through neighboring nodes. Recommendations are weighted according to the direct trust placed in the recommending neighbor, so that not all third-party reports carry equal influence. This provides a route to estimating trust where direct communication history is insufficient, while recognizing that recommendations themselves may be unreliable.
Virtual-force path planning favors trusted sensors
Once trust values are available, the model uses them to influence the mobile collector’s route. The proposed method represents the collection path as a flexible, rope-like curve. Trusted nodes exert virtual attraction on the path, while untrusted nodes exert repulsion. The combined forces shift the path toward more trusted areas and away from less trusted ones.
This approach connects security assessment with the collection task itself. Instead of calculating node trust and then following a route that ignores it, the path-planning process uses trust values to guide which cluster-head locations the collector approaches. The initial route is adjusted iteratively, subject to movement limits on the path. Those limits help keep the route within the specified collection distance and account for constraints in how adjacent path points can move.
In the network structure described, sensors are organized into clusters. Cluster heads evaluate member nodes, and the mobile edge node collects data from trusted cluster heads as it moves. Sensors outside the collector’s direct communication range can relay data through the network’s existing multihop routing approach. The collector therefore combines planned mobility with communication through the clustered sensor network.
What the reported simulations indicate
The study evaluated the approach in simulations, including a setup with 100 sensor nodes deployed across a 300-by-300-meter area and 30 cluster heads. The reported results indicate that virtual-force planning can shift the route toward trusted nodes and away from untrusted ones. Comparisons with other trust-evaluation approaches also reported lower network energy consumption for the proposed method under the tested conditions, with energy use increasing more gradually as the share of malicious nodes rose.
These findings are simulation results, not a guarantee of performance in every deployment. Actual outcomes would depend on network layout, communication conditions, trust-evaluation inputs, movement constraints, and how well the model’s assumptions match the application.
Practical implications for automation teams
For system integrators and operations leaders, the central idea is to make data collection decisions depend on both source credibility and collection cost. A trust score can help distinguish sensors whose readings deserve priority, while an edge-based collector can handle collection without placing the same processing burden on every field node. Route planning then provides a mechanism for translating those assessments into collector movement.
Applying this approach requires attention to how trust indicators are selected and weighted, how recommendations are handled, and how often trust values are updated. Teams also need to consider the network’s clustering and routing design, along with the allowable movement distance and timing requirements for collection. Trust assessment supports data-quality controls, but it does not remove the need to design and maintain the broader security and communications architecture.
Conclusion
Trustworthy data collection links sensor assessment, edge computing, and mobile path planning in one IoT collection model. Multidimensional trust values help identify more and less reliable nodes, and virtual forces use those values to steer a mobile edge collector toward trusted sources. For industrial automation applications that rely on distributed sensing, the practical takeaway is to treat data provenance and collection efficiency as connected design concerns—and to validate any trust-based method against the conditions and constraints of the intended deployment.
Access Official Technical PDF