Industrial IoT systems generate continuous streams of sensor readings, equipment status, and process data. Finding a relevant record across large, heterogeneous datasets can be difficult, particularly when devices have limited resources and operational data must remain confidential. A blockchain IoT data query model offers one approach: keep encrypted data in cloud storage, record compact data indexes on a blockchain, and use hash-based structures to locate records.
The model described here combines blockchain timestamps with Bloom filters and a dual combination Bloom filter (DCOMB). Its central idea is to redirect hash-computation capacity toward data lookup. Rather than placing complete sensor datasets on the blockchain, it uses the chain to organize and verify references to encrypted data stored elsewhere.
Why IoT data needs a different query approach
IoT data has characteristics that distinguish it from many conventional business records. Sensor readings are chronological, data is often appended rather than frequently updated or deleted, and users may be interested in trends across a time interval instead of a single value. At the same time, deployments can produce large volumes of data from systems with different structures and capabilities.
These conditions create practical challenges for industrial data management:
- Heterogeneous data: Device types, data formats, and system structures can differ, making information difficult to use consistently across applications.
- Constrained nodes: Some IoT devices have limited computing, storage, communications, or battery resources. Heavy cryptographic processing can add to those constraints.
- Security and privacy concerns: Distributed, diverse environments can be difficult to secure. Meanwhile, putting sensitive information directly on a transparent blockchain could expose it.
Blockchain can contribute a shared, ordered record with traceable timestamps, but it is not a suitable place for every byte of operational data. A practical design therefore separates the data itself from the information needed to find and verify it.
How the blockchain IoT data query model works
In the proposed arrangement, IoT nodes encrypt their data using a public key and upload the ciphertext to a cloud storage server. They also send compact references to the blockchain: a hash of the data-stream header, a hash of the encrypted stream, and the public key. The blockchain provides an ordered record, while the cloud holds the larger encrypted payload.
This division has two important roles. First, the blockchain timestamp helps associate uploaded data with a period. The model accounts for a delay between block creation and data upload so that the data’s generation time can be estimated from the block timestamp and the specified interval. Second, the hash of the encrypted data can be used to check whether a retrieved cloud copy matches the recorded reference.
When a user requests data, the query process uses the data-stream header hash to locate the corresponding reference and public key. The encrypted data can then be retrieved from cloud storage and decrypted with the corresponding private key, subject to the model’s access and transaction process. Requests spanning multiple time periods may require searches across multiple blocks.
Bloom filters and DCOMB in hash-based query
First, rule out blocks that cannot contain the data
A Bloom filter is a probabilistic data structure that checks whether an item may be present in a set. It uses a bit array and hash functions to map each indexed item to positions in that array. If a required bit is unset, the item is not in the set. If all relevant bits are set, the item may be present.
That distinction matters: Bloom filters can produce false positives, meaning a lookup may indicate a possible match when the item is absent. They do not provide a definitive location or eliminate the need for a more detailed check. Their usefulness is that they can quickly rule out many non-matching blocks with compact indexes.
Then identify the matching entry
The model builds a Bloom filter for the indexes in each block, then uses DCOMB for more specific lookup. DCOMB consists of two layers of combination Bloom filters. In this design, the public key acts as an identifier and the stream-header hash is the index element. The query module can use the input hash to return the associated public key, subject to the possibility of hash-related false positives or classification errors.
In practical terms, the lookup has two stages:
- Check each block’s Bloom filter to determine whether it could contain the requested index.
- For blocks that remain candidates, query DCOMB to identify the relevant entry and return its associated information.
Because the indexes are based on hash operations, the authors propose that mining hardware capable of hash calculation could perform this work. This is a proposed reuse of computational capacity, not a claim that every mining device or deployment will be suitable for every query workload.
What the design means for industrial data systems
For system architects and integrators, the model illustrates how responsibilities can be split across edge devices, blockchain nodes, query processing, and cloud storage. IoT nodes encrypt and submit data; the blockchain retains timestamped references; a processing module constructs and searches the filters; and cloud storage holds the encrypted streams. The design also allows block searches for separate periods to be performed independently, creating the possibility of parallel query processing.
The approach reflects several useful design principles:
- Keep large payloads off-chain: Store encrypted sensor data in cloud storage and use the blockchain for compact references and integrity checks.
- Use time as a query dimension: Organize data around the blockchain’s chronological blocks to support time-bounded retrieval.
- Account for probabilistic results: Bloom-filter and DCOMB lookups can return false positives or classification errors, so query accuracy must be assessed alongside speed and storage needs.
- Match the method to data behavior: The structure assumes data is generally appended, with little need for updates or deletion. Workloads requiring frequent changes may need a different design.
The reported experiments found that DCOMB had better random-read performance and a lower error rate than COMB, and that both filter-based query approaches performed better than MySQL in the tested comparisons. These findings describe the paper’s experimental setup; they should not be treated as a universal benchmark for all industrial databases, hardware, or workloads.
Conclusion: evaluate query structure alongside storage and security
A blockchain IoT data query model can combine chronological records, encrypted cloud storage, and hash-based indexes to address data discovery and integrity in distributed IoT environments. Bloom filters help screen blocks, while DCOMB provides a further lookup stage using hash calculations. For industrial decision-makers, the practical value lies in understanding the trade-offs: compact indexes and potential parallelism must be weighed against probabilistic errors, data-access requirements, and the workload’s update patterns. Any implementation should be evaluated against its own devices, security model, storage architecture, and query demands.
Access Official Technical PDF