6 min read

ROAMING: HOPPING FROM NETWORK TO NETWORK IN REAL-TIME

Featured Image

Authors 

Dr. Christian Dombrowski 

Xavier Bush 

Jesper Lindström

In our series so far, we’ve demonstrated that existing wireless technology can easily meet performance requirements for time-critical industrial applications. That said, hurdles still remain regarding a key Industry 4.0 use case – flexible, reconfigurable shop floors enabled by mobile stations. Can existing methods provide these mobile stations with a stable, real-time connection over their entire operations area?

Currently, the two most popular techniques to ensure wireless coverage are roaming and mesh networks. This entry will analyze both methods and explain why roaming is the ideal solution for industrial automation.

 

Mesh Networks: A Clever Technique that Overuses Wireless Resources

Mesh networks have become increasingly popular in recent years. This is due to both their flexibility and current implementation in Wi-Fi and WirelessHART networks. The concept of mesh networks itself is simple and ingenious - all nodes connect dynamically and non-hierarchically to as many other nodes as possible, then route data to the receiver via the most direct path. Every mobile phone user has at some point found themselves in a room with a weak and constantly dropping Wi-Fi signal, impacting their user experience. Mesh networks provide an effective solution by harnessing other Wi-Fi-capable devices in the vicinity (since there are typically several of these devices in any given location) to create a clear path to the router.

A mesh Wi-Fi network can also increase its coverage area with the number of stations added. A baseline coverage area of 20m can increase to 100m simply by adding five new stations in strategic locations. This solution is highly flexible and versatile as all stations can both send and receive packets.

mesh-1

A mesh network comprised of AGVs, forklifts and a human operator (master)

That said, certain flaws are inherent in mesh networks that make them unsuitable for time-critical automation. To start, mesh network transmission latency can be long and unpredictable – if a transmitter is two “hops” away from a receiver (see figure above), the transmission’s latency is multiplied threefold. The longer it takes for a data packet to transmit, the longer said data packet occupies the channel and prevents other stations from transmitting. This latency increases further with the density of mobile stations in the network. Furthermore, each rerouting attempt requires an update to the network’s routing table, which can result in a lost connection. As cycle times in industrial applications are heavily optimized, mesh networks are not scalable beyond a certain threshold.

Mesh networks also struggle to guarantee constant service, especially when wireless stations are mobile. Consider automated guided vehicles (AGVs), which often operate in large shop floors with many obstacles between them and the central control unit (CCU) - typically a worker with a control tablet. While a high number of AGVs would increase the network’s coverage, all units could still end up in a single area out of range of the CCU, triggering an emergency stop. The CCU-holding worker must then walk over and reactivate them manually. It goes without saying that a setup with this kind of productivity risk is unsuitable for time-critical industrial requirements.

meshfail-2

All mobile stations have accidentally moved out of range of the CCU (the operator)

These two key disadvantages – compounding “hop” latency and unpredictable connections – make mesh networks unsuitable for wireless automation, therefore other methods must be explored.

 

Roaming: A Proven Solution for Wireless Automation

Roaming, by definition, is a signal packet being “handed over” from one network to another. To the average consumer, roaming often refers to mobile phone service beyond the standard coverage area of a service provider, often while travelling abroad or in sparsely populated areas. To provide a seamless user experience, handovers must be imperceptible (typically around 50ms), made possible through strategic placement of signal tower base stations. These signal towers make up what is known as a backbone network - unmoving, predictable wireless stations providing reliable coverage to unpredictable mobile nodes. This same principle can be applied to wireless industrial networks.

Roaming-capable wireless networks have yet to be deployed in factory production processes. While 5G, falling under the Ultra-Reliable Low-Latency Communications (URLLC) umbrella, is developing industrial wireless solutions for this exact purpose, a long wait remains before the necessary infrastructure and retrofits are in place. Thankfully, existing non-cellular wireless technologies (such as WLANs) can easily meet wireless automation’s requirements, with little to no overhauls in existing infrastructure.

 

Full Area Coverage Through Backbone Networks

In the context of wireless automation, a backbone network is comprised of non-mobile wireless access points (routers) connected – usually by cable – to the network CCU. A strategically placed backbone network can easily cover all corners of a shop floor, providing uninterrupted service to all mobile stations (AGVs, untethered robotic arms, autonomous forklifts, etc.). These mobile stations continuously evaluate signal strength of their current and adjacent subnetworks, roaming between subnetworks when necessary to optimize performance.

Roaming_ER-1-3

Two EchoRings (subnetworks) comprising a backbone network

 

Real-Time Roaming

Real-time, low-latency handovers between backbone subnetworks is the second piece of the wireless automation puzzle. To achieve this level of performance, one of two intelligent mechanisms must be implemented: application-driven or network-driven decisions.

Application-driven decisions provide centralized network management control. This means that the network CCU can determine the optimal signal a mobile station can roam to and the optimal time of the handover itself. Handovers are determined based on both the mobile device’s location (detected by an internal or external positioning system) and the network’s deployment plan. This precise control comes at the cost of latency, especially in larger and more complex industrial networks with more nodes. As shown in the OSI (Open Systems Interconnection) graph below, an application-driven handover request to the CCU must pass from the Application Layer (Layer 7) to the Physical Layer (Layer 1) and back again - the entire span of the OSI model. Compounding the issue is that each OSI layer runs on a unique protocol and requires a unique translation.

Handover-Decision-1-4

Network-driven decisions on the other hand see each individual station decide when and to which signal to conduct a handover, independent from the CCU. To achieve this setup, mobile stations analyze, send and receive continuous updates to each other on the relative signal strength of each subnetwork. The key advantage of network-driven decisions is their low latency even in large and complex networks, as the handover request must only pass from the MAC layer (Layer 2) to the Physical Layer (Layer 1).

These requests are also application-transparent, allowing the CCU on the Application Layer to focus solely on its assigned task. The key disadvantage of network-driven decisions is their lower precision (and less effective network optimization), as they lack an application-level overview.

Ultimately, both application and network-drive decisions feature unique strengths and weaknesses and employing one over the other hinges on a use case’s complexity and performance requirements. Regardless, automation use cases involving mobile stations must ensure that all handovers occur within the application’s set latency, to avoid triggering an emergency stop.

EchoRing: Real-Time Roaming

EchoRing was developed from the ground up with wireless automation in mind. It employs a backbone infrastructure to ensure continuous real-time connectivity, and easily adapts to both ad-hoc (decentralized) and classical access (centralized) network topologies. EchoRing also supports both application-driven and network-driven decisions, allowing users to select the method most suited to their application’s requirements. This flexibility allows for optimal functionality in a wide variety of use cases.

As explained above, network planning is key when setting up and deploying backbone infrastructure for an EchoRing network. The user must identify key service areas that require uninterrupted connectivity, then dedicate EchoRing nodes to these areas as necessary to comprise the backbone network. It is also important to determine the required number of network “rings” (subnetworks) and allocate operational channels by order of importance.

Taking all this into account, we can conclude that EchoRing is the technology to use in wireless automation applications.