Duplicate JSON Fields in Telemetry: Reject Ambiguous Payloads Before Mapping
A telemetry message can look reasonable in a log yet give different readings to two consumers. If the same JSON object contains a measurement name twice, the receiving applications may disagree about which value survives. Detect that ambiguity before ordinary mapping loses the evidence, then correct the message producer rather than choosing a convenient reading.

Original conceptual illustration; not a specific product, live software screen or customer installation.
Key takeaways
Duplicate field names within one object are a data-contract problem, not a sensor calibration problem.
Check the original message before a normal parser can discard a repeated member.
Keep distinct channels and repeated observations in explicit structures.
1. Define what counts as a duplicate
This guide concerns two members with the same decoded name inside the same JSON object. It is different from receiving the same observation twice, or two sensors having the same friendly display name. Those need separate identity and delivery rules.
RFC 8259 recommends unique object member names and documents differing receiver behaviour when names repeat. A receiver may retain only the last value, fail, or expose several values. A message that one tool accepts is therefore insufficient evidence of an interoperable contract.
The same name can legitimately appear in separate objects. For example, each item in an array of channel objects may contain a value member. Validation must track object scope rather than banning a word from appearing anywhere in the message.
2. Preserve evidence before conversion
Capture a controlled sample of the original message, its source identifier and its receipt time under your organisation’s access and retention rules. Remove credentials and unnecessary operational details before sharing it. A screenshot of a formatted object may already show only one surviving value.
Ask the integration developer where duplicate-member detection occurs. If detection starts after conversion to a dictionary that keeps only one value, the check may be too late. The chosen parser or validation stage needs to expose repeated members before they are collapsed.
Also test escaped spellings of names. The text pressure and a JSON spelling using a Unicode escape for one of its letters can decode to the same name. Compare decoded names within scope; searching raw text for a repeated literal is not a complete test.
3. A hypothetical pressure example
Suppose the source sends one object containing pressure_kPa with value 420 and, later in that same object, pressure_kPa with value 460. A commissioning spreadsheet and a service log appear to disagree. Recalibrating the transmitter would not resolve the ambiguity in the transmitted structure.
If these are separate channels, the producer should give them distinct identities, such as inlet_pressure_kPa and outlet_pressure_kPa. If they are two observations, use two records with the required time and channel context. Do not average the conflicting fields merely to obtain one acceptable number.
Keep the rejected sample linked to the corrected producer revision. The example is a proposed data-quality workflow, not a claim that every ProSight Software deployment includes a particular parser, quarantine screen or automatic repair function.

Original explanatory diagram. Hypothetical values and proposed checks are described in the text; this is not a product specification or field result.
4. Turn the rule into an acceptance test
Use harmless synthetic messages in a test environment. Include a valid single field, a repeated member, the same member name in separate nested objects, and two observations in an array. Verify the producer, any intermediary transformation and the final consumer together.
Write an expected outcome before running the test: ambiguous objects are rejected or held for investigation according to the agreed integration contract. Record the diagnostic location, the retained evidence and the person responsible for correction. An unexplained missing point is not a useful failure response.
Obtain a sanitised original message and identify each object boundary.
Verify duplicate detection before member values are collapsed.
Run normal, duplicate, nested and escaped-name test cases.
Correct the producer and repeat the full consumer-path test.
5. Common mistakes and limits
Changing key order to make a last-value-wins parser return the preferred value leaves the contract fragile. Renaming a field only in the visualisation can conceal the original ambiguity. Replaying the same malformed message repeatedly adds load without improving its meaning.
Duplicate-name detection does not establish that units, timestamps, channel identities or numeric precision are correct. Treat it as one test in a wider intake contract. Confirm the implementation available for your deployment before specifying operational behaviour.
For the next integration review, bring one raw producer sample and the expected channel map. Agree where ambiguity is detected, how the fault becomes visible and what evidence proves the producer is corrected. That gives commissioning teams a repeatable decision instead of a parser-dependent guess.
Related guide and project resources
Technical references
Need help with your application?
Send the relevant point list, equipment documentation and a sanitised example of the behaviour you need to verify. Confirm the available functions and integration requirements for the exact equipment and deployment.

.png)
Comments