What's the choice of warehouse management software? RFID capture module and hardware adaptation assessment points
Assessing the warehouse management software with RFID collection capability, it is recommended that candidate manufacturers run a real business first: take the labels and readers planned for purchase, complete receipt under representative loading conditions, process differences and inventory submission, and intentionally cut off the web, retransmit them. This process can also expose the radio frequency matching, the reliability of interfaces and the problem of operational rules, which is closer to delivery than showing "how many labels read" on a stand-alone basis.
The selection does not require that WMSs be directly connected to the reader. The device driver can be deployed at the terminal, the edge gateway or RFID intermediate, and WMS is responsible for receiving identified and verified business events. The key is to make clear what responsibility each layer has, which layer can be traced when a malfunction occurs, and which capabilities need to be re-tested after replacement.
I. Determine what labels represent before discussing the compatibility of software
The "Use RFID for Stockpile Management" may mean at least three things: labels bind individual items, labels bind whole boxes, or labels bind revolving trays. The data model of the three is different, costs and acceptance methods. The tray label is read only to show that the tray identity is identified; to confirm how much is on the tray, it also requires a valid line-up and box-to-carrying connection.
It is suggested that a line of business be selected, and that the corresponding "identify the object -- the action -- the result of the transaction" be written. For example: Identification of tray P001, execution of arrival checks under purchase orders, results in a record of receipt pending inspection, rather than immediately increasing available stocks. Then it is clear who maintains the relationship when detached, mixed tow, returned, and who maintains the relationship when changing the standard. Without these agreements, even if the software reads it stable, the software can calculate errors.
There is also a distinction between identification and location confirmation.Handheld reads labels at a certain location, which does not amount to the exact coordinates of the labels that have been detected. If you need confirmation of the library position, you can establish relationships by scanning the coding number, the controlled read area or the validated location determination rules; if you want to keep positioning, you should separately evaluate the locator technique, the way and the precision of the position.
`Supporting RFID' to three-layer adaptation requirements
| Level | Capacity to be validated | Evidence to be obtained at delivery |
|---|---|---|
| Device & Driver Layer | Connect to specified models, start and stop inventory, configure antennas and power, receive label reports, identify breakers | Reader type, solid, SDK or protocol version, and actual call record |
| Collection and Rules Layer | Reads by job window, filters duplicate and cross-area readings, links jobs and save anomalies | Correlation between original readings, rule versions, candidate events and causes of anomalies |
| WMS Business Level | Checking documents, cargo owners, batches and status, processing discrepancies and controlling inventory services | Task records, inventory changes, interface validation and operational audit |
The device drive library reduces the amount of development, but compatible lists must be specific to models, solids and functions. Capable of inventory, does not mean that you can write labels, calls GPS, uses external triggers or restores a breakout. When the manufacturer says "the brand is aligned", it should continue to check whether these functions have been tested.
LLRP is a low-level interface between the reader and client, which can be used as an option for access to equipment. It does not define the procurement receipt, stock freeze or batch allocation rules, nor can it prove that any equipment is immune from development replacement by supporting LLRP alone. The review should confirm the version supported by both parties, information, extension of capacity and operation; the equipment user SDK can also deliver a reliable solution if the interface is clearly responsible and maintained.[1]
III. PUT THE LABEL, AIR AND TRANSPORT FROM THE PLACE OF THE ASSESSMENT
Hardware fit cannot be the maximum reading distance. What is needed on site is " a target identified reliably in the given action, while not being read to the next one as much as possible." For access, reading further may bring goods from the adjoining area into the current task; higher power does not guarantee that the problem posed by polarization or loading structures disappears.
When you test a sample, you should fix the label and adhesion position at the same time, the material to be marked, the packaging form, the stacking method, the antenna position, the equipment parameters and speed of passage. The effect of metal and liquid should be measured on site, and it is not appropriate to apply a common percentage of the loss from distance. Anti-metallic labels should also verify the installation surface, the spacing conditions and the actual direction, and cannot read only by name.
The channel collection should also indicate when the operation begins, when it ends, how to determine direction. It can combine photo-power triggers, door controls, mission startup or multiple area observations, but specific programs are validated by field action. The antenna number in the label report, time and signal strength, is analysed to support a single intensity threshold that does not guarantee any distinction between adjacent reservoir areas.
- Fixed equipment:PoE requires equipment to match the standards and power budget of the end of the supply, and cannot only confirm “access to the net”.
- Handheld terminal:Checks of operating systems compatible with SDK, offline missions, swipe coordination, battery replacement and cryogenic conditions.
- Cycle label:Check for fixation, readability after cleaning or collision and for reloading of the vehicle.
- One-time label:Checks print codes, first-reading checks, bad mark segregation and duplicate code disposal to avoid the introduction of coding errors into the business system.
Interface acceptance and inspection should cover “no reply after success”.
Once the collection module is submitted to WMS, there may be a failure that could be easily missed: the WMS has already submitted inventory services, but the response was lost in the network, and the capture module considers it unsuccessful and retrys. If the system is heavy only by short-term labels without a permanent operational identity, retesting may increase or reduce the stock again.
Interface design should identify the event identity, task or document number, label and binding object, device and reading point, time of occurrence, time of receipt and status. Which fields are filled in depending on the process; core requirements are to be able to track an operation and recognize a retransmittance of the same operation. The same tray is valid to receive again the next day, it is new, and cannot be filtered permanently because the label numbers are identical.
Also specify the offline cache capacity, how to last, restart, re-route and fail to retry. When original files are cancelled during offline, the back-up event should recheck the current business status and not execute the old command without conditions. The device is online, messages have been sent, WMS has received and inventory has been submitted, which should be separately checked in the surveillance.
V. Acceptance and acceptance of a recalculated set of data, rather than an integrated accuracy rate
Start with a controlled test: there are 100 target tags, and the reader picks up 98 of them plus two non-target tags from an adjacent area. Looking only at the total of 100 reads would wrongly suggest a perfect result. Using sets of unique tags, target read coverage is 98 ÷ 100 = 98%, and the non-target share is 2 ÷ 100 = 2%. These figures illustrate the method, not an industry pass threshold.
Next, check the business logic. Were the two non-target tags filtered out? Were the two missed reads sent for rescanning or review? Did the system still automatically post a complete receipt before discrepancies were confirmed? Keep physical read metrics separate from transaction posting metrics. Manual entries can hide collection problems, and repeated reads can be mistaken for high throughput.
| Item | Test Action | Check results |
|---|---|---|
| Read Borders | Place known labels in the neighbourhood at the time of passage of target goods; test each to be full, different orientation and speed | Read out of the list of independent targets and non-targets, save original reports |
| Duplication and regression | Stay, back, re-pass and repeat the same event | Observation of the number of potential incidents and inventory services, confirming that no duplication of submission will be made |
| Network & Restart | Collecting disconnected network, submitting it and discarding the receipt and restarting the collection process | Cached events recover, disposal results are searchable and inventory only changes should be made |
| Operational conflicts | Add unknown tags, wrong batches, freeze objects or cancel original task to replace | Enter into a clear refusal or review process, and retain the cause and handler |
| Maintenance replacement | Replace the same-specified equipment after backup configuration, restore parameters and rerun critical examples | Configure recoverable, correctly mapd and compatible conclusions can be verified over time |
The acceptance report shall indicate the number of tests, each loading condition, a valid sample and a failed sample, recording the results after the first adoption and review, respectively. Time indicators may be broken down to collect, transmit, verify and perform the transaction time-consuming, indicating the starting and ending of the time and the time conditions for the equipment. The eligible line is determined by the operational risk and the operating rhythm, and it is not appropriate to replicate the percentage of other warehouses.
VI. PROCUREMENT BORDER BETWEEN MAINTEN RESPONSIBILITY AND CHANGE COSTS
A comparable quote should be given with labels and consumables, equipment, installation, drive, collection rules, WMS interfaces, authorization, training and ongoing maintenance. The label is used as a single item consumption or vehicle recycling, and the cost is calculated on an annual basis or on a swing life; it cannot be estimated by "one label for each cargo".
The responsibility for this is to: who returns to the test after the equipment solids are upgraded, who maintains the interface after WMS upgrades, who re-checks the labels and re-profiling the site, and who saves the original data when the failure occurs. If the rules can only be changed by the manufacturer, then also indicate response time, how the fee is calculated and how the export capacity is configured. Let the supplier offer a quote on the same set of tasks, the abnormal and the receipt/receipt material, to compare the real delivery range.
The final selection sequence can be: Define the object and business actions, jointly identify the software interfaces and hardware candidates, pilot under representative working conditions, then evaluate them by evidence. The value of the RFID module is reflected in the fact that each business can be explained, failure can be restored, equipment can be maintained; these capabilities need to be demonstrated together with on-site process and interface testing.
References
[1] GS1, Low Level Reader Protocol (LLRP) 1.1.0. The terms of reference for checking the reader ' s interface with the client; specific equipment capabilities still need to be validated on the basis of the manufacturer ' s realization and testing.View official standards。

Ms. Lei
Mr. Pang