How to build an asset management system based on RFID? Hardware selection and architecture
What hardware does the RFID asset management system typically require?
A running RFID (radio frequency recognition) asset management system, the hardware dimension usually consists of four parts: electronic tags, readers, antennas (fixed reading and writing machines require an antenna, handheld terminals with antennas), and intermediate servers or edge gateways. Labels are placed on assets, readers are responsible for sending radio-frequency signals to read label data, middle items are used to clean raw data that are read, filter, link to the upper application system.
In terms of choice, you have three questions to answer: what the asset is, in what environment, and how much grain is. Once the answer is clear, the hardware configuration will be basically established.
How do we select RFID frequencies? Where are the low-frequency, high-frequency and ultra-high frequency differentials?
The frequency RFID uses is three-tier: low frequency (LF, 125-134kHz), high frequency (HF, 13.56 MHz), ultra-high frequency (UHF, 860-960MHz). These three sets vary significantly in the asset inventory scene.
Low-frequency penetrators are strong, but read at short distances (usually a few centimetres) and suitable for close range identification in metal environments such as tool management, subwell equipment marking.
High frequency reading distances are usually within a metre, metal resistance is generally the most used for people's doors and IC card scenes, and relatively little for asset counts.
UHF is the first choice for asset stocktaking. Read range can be adjusted from 1 meter to more than 10 meters. With directional antennas, it can read dozens or hundreds of tags in one pass over long distances. Metal and liquids affect it substantially, so use specialized on-metal tags or leave a gap between the tag and the metal surface.
In practice, the scenario of a large number of metal assets (e.g. room equipment, hardware moulds) usually selects UHF Zon-metal tag; liquid environments (e.g. reagent bottles, liquid bags) select the UFF flexible anti-liquid label or downgraded HFC option.
How much is the label cost different?
The normal URHF non-dry glue label (paper) unit price is usually in the range of $0.3-1; the UHF anti-metallic labels are in the order of $2-8, HF tag 0.5-3; low frequency labels are high because they are small in use. When assets are large, the label costs are large in the budget, and are multiplied by the full amount of assets.
How do you match a fixed-reader with a handheld terminal?
RFID readers are divided into two main categories: fixed and handheld. The roles of the two are different in the system.
Fixed-type readers are usually deployed at the entrance/entry, through the channel, and at the end of the shelf, which is associated with antennas to form "reading doors" or "reading channels." Its role is to record the assets in and out of the room — for example, when equipment moves through the door, the system automatically records the time and direction of departure. The selected type of fixed-form readers depends on the number of antennas (4 antennas, 8 antenna types are common), the agreements supported (EPC Class 1 Gen2 is industry standard) and whether it supports the supply of power (facilitating wire) by the PEE network.
Handheld terminals are the main force of asset counts. They are carried by inspectors who move around in warehouses or aircraft rooms, scan the shelf. The handheld terminal selection point is: whether to carry RFID modules, whether to scan with bar code (many assets are redundant with RFID and barcodes), protection level (over IP54 more than better for industrial environments), and re-opening (one charge at least one shift).
A system usually has a fixed and handheld combination. Fixed-based solutions "Auto-Cut in and out" and hand-held solutions "Systems on a regular basis." Only hand-held can run, but access data is manually recorded, with much more work to do.
How should the system structure be designed?
The RFID asset management system is typically structured on four levels: label, collection, middle, application.
The label layer is the electronic label on the asset, which stores the asset code (EPC area), user-defined data.
The collection layer includes fixed-form readers, handheld terminals, door-barrel antennas and is responsible for the conversion of radio frequency signals into data.
The middle layer is the brain of the system. It does several things: filters duplicate data (the same label has been read many times over), links to the reader and antenna position, puts time and position stamp on raw data, and pushes the data to the upper layer. The intermediate can be placed on local servers or on edge gateways.
The application level is used by the operational personnel and includes functional modules such as asset desk accounts, inventory tasks, access return, depreciation at end of life, report analysis, etc. This level is usually integrated with the enterprise’s existing ERP, OA, financial systems.
The data flow is about the same: the label is read and poignant, the reader sends the data via Wi-Fi or a web line to the middle medium poignant, then washes it and pushs it to the application layer poaching, and triggers the process.
What do we have to look at in the network and deployment?
Fixed-type readers usually follow the TCP/IP cable network, deployed in weak electrical rooms or cabinets. Handheld terminals go Wi-Fi, warehouses or rooms need to be pre-wire-covered, and leaks occur in poorly covered corners.
If assets are distributed in multiple plants, the "Specific capture + centralized management" structure can be used, with local intermediates deployed to each plant area for uniform reporting to the Headquarters platform. This will ensure real-time and cross-geographical accessibility of field collection.
What are the common pits in the construction process?
Pitfall 1: Choosing tags without considering asset materials.The labels on the metal surface must be anti-metal, and general labels are largely unreadable.
Pipe II: Antenna deployment is imaginary.Read-and-write distance is not a straight line of distance, metal reflection, antenna angle, and labels are going to affect the actual reading effect. Important points (e.g., door doors to the machine room) must be tested on site, read over and over again with real assets, and then fixed after confirmation of reading rates.
Pit three: The middle is not heavy.The same tag can be read 20–50 times if it stays in the portal for three seconds. Without deduplication, duplicate entry and exit records will flood the system and make the register unusable.
Pipe IV: Neglect and docking of existing systems.Many enterprises already have ERP or financial systems, and there is a set of asset data. If RFID is only operating independently, without access to ERP, it's half the value. It starts with planning for data interfaces (usually RET API or database middle tables).
Pitfall 5: Overlooking tag printing and initialization.Tens of thousands of labels need to be coded, initialize the EPC, and match the asset desk account. This step is a lot of work, and it's better to start bulk processing one month before the system goes online.
How long does it take to get on line?
A midsize project covering 10,000-50,000 assets at one or two plants typically takes three to six months from requirements analysis to go-live. Most of the time goes into three phases: site survey and hardware selection (2-4 weeks), system deployment and integration testing (4-8 weeks), and trial operation and optimization (2-4 weeks). Doubling the asset count or adding plants lengthens the schedule accordingly.
If you're planning a RFID asset management system, you might suggest a small pilot -- pick a warehouse or room, run the entire process from tagging to inventory, put hardware selections, antenna deployments, and all the pits that connect the system are stepped on first, then spread them out on a large scale. Shanghai-based Siexi technology has done this in several industries, usually putting down a high return rate for official deployment.
Finally, RFID is not a panacea. For the middle of a dense metal shelf or for the hidden assets, the best RFID program can't read. This point goes with bar-code scanning or hand-counting, so don't expect a single technology to solve all the problems.

Ms. Lei
Mr. Pang