Improving industrial wireless reliability with PRP

Wireless solutions are often thought to be tainted by susceptibility to interference and lack of reliability. Despite recent technical improvements, wireless transmission technology still elicits concern when mission-critical processes have to be controlled or monitored via WLAN connections. However, by using standards-based redundancy techniques, such as Parallel Redundancy Protocol (PRP), concerns about reliability and availability can be greatly alleviated. This article describes the basic technical principles, requirements and application scenarios for the application of PRP in WLAN networks.

Regardless of industry concerns, wireless LANs have become an enabler for many of today’s communication applications in industry. WLAN is an excellent solution whenever using cables would be too heavy, unreliable (wear and tear), costly or simply impossible due to moving parts and vehicles. In addition, the use of wireless on the production floor enables a completely new approach for planning and executing production processes (Industry 4.0).

The ongoing technical advances and much wider acceptance of wireless solutions in recent years is also fostering increasingly challenging and highly sophisticated application scenarios. Yet the reliability and quality of service of wireless connections can cause concern for applications with strict reliability and latency requirements.

Examples of such critical or sensitive applications include controlling production workflows or video systems with safety tasks, such as the monitoring of hazardous locations in a plant or monitoring the interior of passenger rail cars. Network interruptions can quickly lead to serious problems and consequently high follow-on costs.

PRP - packet redundancy

To counter network disruptions in wired industrial Ethernet networks, redundancy techniques have been established to ensure that the network continues to operate smoothly even if individual connections fail. These redundancy techniques can also be used in wireless networks in order to significantly increase the reliability and robustness of the connections.

One key example is standardised Parallel Redundancy Protocol (PRP). IEC62439 PRP is used increasingly in wired environments to enable seamless redundancy or loss-free switching in the event of failure in a network path or a device. To achieve this, data packets are duplicated and transmitted in parallel across two different and independent network paths. Before the duplicated packets are delivered beyond these network paths, the parallel streams are merged and duplicate packets are removed. If a single path fails, packets from the other path will be used. The application relying on this network can therefore continue to work without failure despite serious disruptions in the network. (Figure 1 shows PRP in operation in a wireless scenario.)

Figure 1: PRP over two WLAN transmission paths - the redundant transmission compensates for packet losses and counterbalances load and interference-related transit time differences.

PRP can also be used in a wireless environment, although the impact manifests itself in a completely different and even more beneficial way from in a wired scenario. This is because parallel redundancy can be used not only for total network disruptions, but also to compensate for the inherent small-scale disruptions (such as interference) in a wireless network. When PRP transmits packets simultaneously on two different wireless transmission paths (Figure 3), the effects of individual path packet losses can be eliminated; a transmission fault or receive error on a path only becomes visible if both paths fail simultaneously for the exact same packet. In other words, uncorrelated packet losses are never seen by applications employing PRP techniques.

Although the mechanisms used by PRP are the same in both wireless and wired scenarios (packet duplication and elimination), the effect achieved is more dramatic for wireless. While use of the PRP allows a seamless switchover between two networks in both a wired and wireless scenario, its use in a wireless scenario immediately offers a number of additional advantages:

1. Compensation for individual packet losses in case of temporary disturbances, such as interference caused by other radio systems, increasing reliability dramatically.
2. Decreased latency, since the faster of the two duplicated packets is always forwarded.
3. Reduced transit time fluctuations (jitter); and long delays, caused by an occupied medium or by network layer retransmissions, are reduced because fluctuations only appear if both packets arrive late.

Benefits in practice The benefits of deploying standards-based PRP can be demonstrated using a simple example: Assuming the loss rate is identical on both paths and is approximately 0.1%, the loss rate (or loss probability) for the overall PRP system would be just 0.0001% (0.001 x 0.001 = 0.000001) - an improvement by a factor 1000. This calculation assumes that losses are evenly distributed and are not correlated. To achieve this in practice, it is necessary to ensure there are no influencing factors that would impact both radio channels equally, a best practice in building reliability. A key example of this diversity is operating both paths in different frequency bands. As a result, a competing radio transmission or other environmental influences cannot affect both paths at the same time. Other factors that cause correlated losses and reduce the uniformity of the loss distribution should also be minimised. For example, permanent overloading of a connection can cause sequences of packets to be dropped, which drives up the loss rates for this connection and therefore significantly worsens the combined loss rate at the same time. These dramatic improvements can also be achieved in reality. In practical tests, the perceptible packet loss for the application with the PRP was reduced from 0.105% and 0.101% for the individual connections to 0.00021% using a parallel redundant PRP connection - an approximately 500-fold improvement. Another positive effect of the use of PRP is that the network latency and transit time differences - jitter - decrease significantly in the network. A reduction in the average latency from 3.1 or 2.8 ms to 1.7 ms can be observed in practice in the above example. The jitter value likewise falls from 0.45 to 0.23 ms. The reason for the improvement in these metrics is that the faster of the two packets transmitted across the wireless links is always forwarded with the PRP.

Figure 2: PRP over two WLAN transmission paths - improvements in average packet loss, jitter and latency can be dramatic.

Outlier packets with long transmission times, as are known to occasionally occur with WLAN because of the shared medium and the non-deterministic channel access, can be largely eliminated in this way. Consequently, three of the most important quality indicators for a network (loss rate, jitter and transmission time) are improved significantly with PRP.

Topologies and applications

While PRP represents a significant improvement in the redundant protection of individual transmission paths as outlined above, the fact that it is not limited to pure wireless transmission makes the flexibility of this standardised solution all the more obvious when it comes to complex network structures. Even though proprietary WLAN redundancy solutions potentially offer enhanced transmission performance, such improvements are always focused on an individual transmission path. Standardised PRP on the other hand allows more complex scenarios to be realised with wired and wireless Ethernet connections, as well as mobile applications with roaming devices. In a scenario in which PRP is used over both a wired and a wireless transmission path, the wireless transmission path can therefore be used as a switchover-free backup connection for the wired path in the case of applications with difficult constraints (such as moving parts or high temperatures). Such a combination is not possible if proprietary WLAN redundancy solutions are used.

Figure 3 shows the use of PRP in a mobile scenario: a dual-radio client (on a moving machine or a vehicle) travels along a path with several access points. The client can operate two connections at the same time, which means that the path can be protected with PRP.

Figure 3: PRP in a WLAN network comprising several access points and a client. Elimination is performed at a central location with a PRP-enabled switch.

The client can also establish the redundant connections to different access points along the route and roam from access point to access point with one of the two PRP connections remaining active at all times. Moreover, the resulting quality of the connection will always be as good or better than the best of the two connections, regardless of mobility effects (such as bad signal-to-noise ratio or fading) since the PRP algorithm automatically chooses the packets of the better link. This allows roaming interruptions and service degradation to be avoided with no switchovers. What is important in this scenario again is that PRP is not limited to the wireless channel, since various WLAN connections run over several access points connected to the network in different ways. Duplicate packets must be eliminated at a central point in the network, something that is only possible using a standardised and WLAN-independent method.

More about the PRP standard

Like other well-known redundancy protocols (such as STP and RSTP), PRP is a form of Layer 2 redundancy. In PRP, however, each PRP node has two interfaces with the same MAC address and IP address (at Layer 3) - which allows higher layer protocols to operate over PRP without modification. All the redundancy occurs at the Ethernet frame level. It therefore depends on hardware that supports such a configuration - such as the Hirschmann OpenBAT series of access points.

Non-PRP nodes can be attached to one of the networks only, or can be attached to both through a special device known as a RedBox.

The original standard IEC 62439-3 (2010) is sometimes referred to as PRP-0. More recently it was amended to align PRP with the High-availability Seamless Redundancy (HSR) protocol, which uses a ring topology instead of parallel networks. To achieve this, the original PRP was modified at the cost of a loss of compatibility with the PRP 2010 version. The revised standard IEC 62439-3 (2012), referred to as PRP-1, describes both HSR and PRP. Many technical details are now aligned with HSR, which eases the implementation of multiprotocol nodes. In particular, a redundant transition between HSR and PRP networks is now possible.

Conclusion

As a standardised redundancy solution, PRP is ideally suited to dramatically improving reliability and the quality of service of wireless connections. In addition, PRP allows a variety of network topologies comprising wired and wireless connections to be protected. As a result, loss- and latency-sensitive applications can successfully be operated over wireless connections.