Wireless networks, particularly wireless sensor networks (WSN), extend the reach of sensor networks to locations that were previously not economically feasible. An argument can be made that Industrial Internet of Things (IIoT) devices are an extension of WSN, with the increasing capabilities of 5G and soon 6G.
Eventually, wireless signals must transition to physical connections and associated antenna, wire traces, processors and interconnecting cables. The laws of physics tell us every transition and associated connection will result in some losses. Connector loss of 0.1-0.5 dB per mated pair is typically the same order of magnitude as losses in the cable connection. It’s a reason to minimize connections and distance between radio and antenna. As designers, we do our best to minimize and compensate for losses through good design practices, signal amplification and antenna design.
Losses at the transition point have two sources: the efficiency and capacity of the wireless input, and the processing capabilities of the gateway device bridging the two mediums.
The bridge or gateway must perform several crucial tasks:
- Demodulation/modulation: convert the radio signal into raw digital bits and vice-versa.
- Error correction and retransmissions: handle lost or corrupt wireless packets—more common in wireless than wired—which includes cyclic redundancy checks (CRC), acknowledging good packets, and requesting retransmissions of bad ones.
- Protocol translation and encapsulation: translate wireless frames into Ethernet's 802.3 frames—and vice-versa—which means adding headers and footers and potentially dealing with different MAC addressing schemes or network segmentations if the device is also used for routing.
Receiving data in one protocol, parsing it, performing translation logic, buffering, and re-transmitting it in another protocol takes time. Depending on the complexity of the translation, the gateway's processing power and the data volume gateway latency typically range from milliseconds to tens of milliseconds. If the processing time isn’t consistent, for example, because the device is running at its operating limits, and the latency varies significantly, it introduces jitter into the signal.
Get your subscription to Control's tri-weekly newsletter.
Generally, any physical limitations in the wireless-to-wired transition are small compared to the differences between the WSN capacity and the wired network (normally 100 MB or higher). The other source of errors when using a gateway, especially if the protocol changes, is the risk of introducing registry-mapping errors.
Mapping errors aren’t only about flipping bits, but also involve the meaning of the data lost or misinterpreted during translation. This can happen due to:
- Mapping data types: forgetting to account for endianness (byte order), different integer sizes and floating-point representations. If the scale or decimal point isn't handled correctly, a value of 10.5 in one protocol might become 1,050 or 10 in another.
- Truncation and overflow: if a value from a larger data type is mapped to a smaller data type in the target protocol, data can be truncated, or an overflow can occur, leading to incorrect readings.
- Scaling or unit conversion issues: data values must be scaled or converted between units (e.g., Celsius to Fahrenheit, PSI to Bar). In some cases, rounding in these calculations leads to incorrect data.
- Incorrect register or address mapping: the gateway maps the wrong data point from the source to the destination. For example, reading the temperature from register PV1 in a HART device and mapping it to Modbus 40010 instead of Modbus 40001. Depending on how the Modbus memory is configured, this also leads to associating an incorrect unit (PSI rather than oF) with the measurement.
- Loss of context or metadata: some protocols carry rich metadata (timestamp, quality flags, engineering units, alarm states). If the target protocol or the gateway doesn't support or correctly translate the metadata, context can be lost, making the data less useful.
Fortunately, proper testing should catch all these errors before the gateway is configured in the field. However, if field changes are made, it’s often more difficult to do the same level of testing as during the initial installation. If not caught, it can have cascade effects, making it even more difficult to find.
Assuming we get the design correct and match the correct gateway to the WSN, the mostly likely source of error is human—just adding meaning to the phrase “lost in transition.”
