How do you design the RFID hardware for the storage management system of wms?
Designing a synergetic structure for WMS and RFID hardware, first of all, to avoid explaining "labels read" as "stock changes." Readers produce label observations at certain times and within the reading area; WMS processes business matters related to documents, owner, batch, storage position and stock status. The two levels require a layer that controls capture, links tasks, handles duplication and preserves the incident logic.
The actual system can be achieved using different devices, such as SDK, standard reader protocol or edge collection service. The structure design is not focused on a component name but on each layer of responsibility, the data boundary and failure recovery path are clear. This paper describes the design approach by device access, margin event treatment and WMS business processing.
First, define business events and then determine what the interface is.
The same label may appear dozens of times over a fixed antenna. These records do not necessarily represent dozens of business actions, nor can they determine how far an object is from an antenna by the number of times read. RSSI is a signal-receiving indicator that is subject to label orientation, mask, antenna, environmental reflection and equipment parameters, which are suitable for supporting analysis in controlled tests, and cannot be used as a basis for determining common ranging results or reservoirs alone.
First, write on-site operations as an event, such as "P-08 is expected to be verified at the receiving point in a receiving mission R-102." The event should allow for associated tag identity, current tasks, reading points, observation time, collecting equipment and judgement. Only after operational verification and submission conditions are met can the system convert its candidate observations into receipt or inventory services.
Also define the status of treatments such as “observing” “candidate matching” “received” “received” and “remitted”. Thus, when equipment is read successfully but no document exists, the system can retain observations and enter into processing rather than quietly changing stocks or discarding anomalies.
Device Layer: Create a sustainable readable boundary
Readers and antennas are responsible for tag communication, equipment status and original reading reports. Deployment design should be based on the operational point and cargo pattern, rather than first using fixed antenna overlap ratios, frequency differentials or reading rates. Metals, liquids, multi-layer packaging, labeling surfaces and speed of movement affect reading results, which must be validated under representative field conditions.
- Defines the reading point:(b) Sets a stable logical mark for the receiving, exit, inventory or workstation and establishes a configuration map of the reader, antenna and logical position.
- Task Distinction Window:Defines when a single collection begins and is completed by mission activation, PV trigger, door control or clear manual operation. The specific trigger should be matched with the process in the field.
- Recording device context:Saves the device number, antenna number, solidware and key parameter versions to allow for comparison of pre- and post-adjustment reading performance.
- Processing equipment anomaly:Monitor break-up, antenna anomaly, clock deviation and cache, and indicate whether equipment is retrofitted and how to avoid the misperformed performance of old tasks.
Where the protocol capability is required, the manufacturer-supported model, solidware, message and range of operations should be verified on a case-by-case basis. GS1 LLRP defines an interface between RFID readers and clients that can be used to control and manage reading and writing; it does not define WMS receipt, inventory, batch or warehouse rules per se. LLRP cannot automatically prove that different brands of equipment can be replaced without any difference, and interoperability needs to be validated through equipment and business examples.[1]
Marginal or intermediate layer: Generate traceable events from reading reports
The acquisition service should keep the original log and generate business candidate events on it. The heavy-thrower goal is to compress duplicate reports in a mission observation window, not permanently shield an EPC. When a label is again valid for another task, it should allow new candidate events to occur.
Event aggregation requires clear rule versions and boundaries, such as observation windows, target reading points, current tasks and permitted labels. Signal intensity can be part of the screening or diagnostic conditions, but if only fixed RSSI thresholds are used to distinguish areas, environmental error may occur after a change; boundary judgement should be combined with antenna setting, trigger signal, mission context and results.
It is recommended to create a stable operational identifier for each candidate event, and to keep the original log reference, process the rule version, first observation time, event generation time and dispatch status. The event sent with a reliable queue or a restored local lasting cache, avoiding a restart of the process, and network flash breaks that lead to untraceable access to the readable data. The queue backlog, duplicate delivery and clock deviation should also be included in the monitoring.
WMS Level: Checking the business context and submitting inventory
When WMS receives a candidate event, the label should be checked for binding, material, owner, batch, unit of measure, storage and stock status according to current valid documents. The tray tag needs the correct aggregation relationship with the cargo in the box; if the tray is split or reloaded, the link records that are valid at the time should be searched and the historical operation cannot be overridden by the current relation.
The inventory submission requires processing, for example, of a tact. WMS may have successfully submitted the transaction after the collection end has sent it out, but the receipt is lost in the network. When reissuing the same business operation, you should return to the original process and avoid repeating the records; and the following action on the same item in another legal task order should be marked with new operations. Re-renewing the legal flow by simply using a label number would prevent the entire operation.
Interface confirmation should distinguish between " accepted " and " completed business processing. Response may include operational markings, end state, reasons for refusal or review, and searchable service numbers. Unverified labels should not be discarded silently, nor should they be directly transformed into usable stocks; original evidence should be retained and entered into disposalable abnormal processes.
Event model to express what happened, not just the label number.
The answer is at least a cross-system retrospective: which object, when the business action occurs, by which reading point or device, which task or document belongs to, and what status it is after processing. The time of incident and the time of system recording should be kept separately, especially when the peripheral device is offline to reconnect.
GS1 EPCIS organises events around objects, time, location, business context and state dimensions, and distinguishes between reading points and operational locations. It can be used as a reference for multi-system event symmetrical design to help WMS, equipment platform and supply chain systems understand the same event; using this reference model does not mean that the project must deploy an entire EPCIS repository or query service.[2]
The event record should also be linked to the correction and manual processing. If you read it and are judged wrong, keep the event, correct the cause and then follow up, it will help explain why inventory changes. Directly overcoming the old state of the database would render audit and variance analysis unfounded.
Acceptance and inspection examples of interfaces and hardware connection
| Test scene | The system should perform | Checking evidence |
|---|---|---|
| The same tab is read in a task window | Generates a rule-compliant candidate, does not repeat the same inventory | Original report, consolidated version of rule, operational identifier and inventory flow |
| Tabs in adjacent areas that are not current documents appear | Identification as non-target observation or pending review, not commingled with current receipt results | List of mission objectives, points of access, non-target labels and reasons for processing |
| Inventory services were successful but interfaces were lost | Return to original results after retesting without double additions or decreases | Request identification, processing status, service number and retry log |
| Collecting continued during network interruptions and re-entry | Retain the date of occurrence, re-check the status and rules of the mission and perform the cancelled task without blindly | Equipment clocks, time of occurrence and receipt, cache order and review records |
| Labels are not bound or box-to-ship has changed | Stop automatic check and enter a clear anomaly process | Tie history, box-to-carry version, operator and disposal records |
| Replace the reader solid or adjust antenna parameters | Implement the case of a return to association, confirming that operational boundaries and incident statistics are still in line | Change orders, configuration snapshots, back-to-back control data and regression conclusions |
The combined results should be read, incident management and business submission of tiered statistics. For example, separate logout of target labels with non-target reading, delayed candidate event generation, number of WMS rejections or pending review, stock transaction duplicates and interface recovery times. Each ratio must indicate sample range and denominator; common fixed accuracy threshold is not a substitute for contract engagement and field baseline.
Order of decision-making when architecture is set
- First, it is determined whether the business actions to be automated are to be automated and whether the identification is for singles, boxes or trays.
- The equipment and trigger are selected by the field boundary, and the acquisition conditions are validated with representative cargo.
- Defines candidate events, mission correlation, abnormal status and inventory submission conditions.
- Cache, retest, tatter, etc., query and correction, and re-drive the grid and restart the service.
- The same set of routine receiving and inspection equipment, collection services and WMS are available for reproduction and the equipment parameters and rule versions are recorded.
The core of the RFID hardware synergetic architecture is to reliably transform radio frequency observations into context-based, reviewable and recoverable business events. Interface protocols are for equipment communications, incident layers for evidence-gathering and mission-related, WMS for operational verification and inventory. Once the border is clear, hardware upgrades, software extensions and problem positioning are clearly based.
References
[1] GS1, Low Lever Reader Protocol (LLRP) standard. Defines the interface between a reader and his client and the ability to control a reader.View official standards。
[2] GS1, EPCIS Standard 2.0.1. Data models describing visual events and information on time, location and operational context.View official standards。

Ms. Lei
Mr. Pang