Dispelling RFID Network Performance Concerns

By Michael Crane

In this case, the RFID data must have priority over other network traffic so it can be communicated without delay to all components involved in routing the material. These priorities are assigned through network quality-of-service (QoS) software, commonly used for high-priority applications such as voice over IP (VoIP).

In addition to high-speed communications within a facility, you should consider the impact of RFID communication over the lower-speed WAN links connecting remote facilities. Large enterprises such as retail chains often connect stores through a fractional T1/E1 WAN, dial-up connection or other low-bandwidth service because these technologies are far less expensive than high-speed, dedicated links. The performance of these lower-speed connections can degrade under excessive loads.

You can maintain WAN performance in a large, data-intensive RFID implementation by using middleware to aggregate and filter the RFID data within the network edge. Filtering reduces the volume of raw tag-read data transmitted within the site and over the WAN by limiting transmission to essential data. For example, filtering can prevent duplicate tag reads from being sent when a package remains in the same read field for an extended period of time.

Network Optimization
Maximizing RFID application efficiency via network optimization is best accomplished through collaboration between the RFID application and network engineering teams. By working together, the two groups can maximize the value RFID will bring to the enterprise architecture, while preserving the performance of network-dependent resources such as SAN and VoIP.

For example, network administrators can use data-flow analysis to optimize the network for RFID before the system is fully deployed. Data-flow analysis requires knowledge of protocol overheads, network forwarding paths and network QoS capabilities because the tag data is actually a very small portion of the traffic generated by a tag-read event. Many RFID readers use XML over TCP to transmit tag data to the middleware. The TCP protocol uses a three-way handshake to establish a connection between devices before sending data; it uses a similar process for session completion. This handshake adds several packets to the actual tag data. Once the TCP connection is established, the tag data is encapsulated within an XML format that adds even more overhead to what started out as a minimal amount of RFID data.

The data-flow analysis lets the network team simulate reader, middleware and application transactions to understand the potential amount of data generated by the RFID network. Through careful evaluation of each read point type within the RFID network and its interaction with middleware and other enterprise applications, the team can begin to project how much traffic the RFID implementation will introduce, and adjust the network foundation to maximize application performance.

Data-flow analysis is especially important in wireless networks, which typically have less usable bandwidth than wired networks. Manufacturing operations requiring RFID commonly deploy an IT infrastructure over a combined wired and wireless network. Network teams can maximize performance over wireless networks by filtering RFID data, and by using high-performance wireless standards such as 802.11a and 802.11g.