The transition to open standard IP switching based on commercial hardware and software has revolutionised the way TETRA networks are implemented, configured and operated, as James Atkinson finds out

The transition to open standard IP switching based on commercial hardware and software has revolutionised the way TETRA networks are implemented, configured and operated, as James Atkinson finds out


Switching technology lies at the heart of wireless radio networks. They control the basic functions of switching voice and short data service (SDS) messaging, status, and packet data calls to and from radios.

Switches hold the databases and network management information on authorised radios and the services assigned to them. They also configure the talk groups, determine the task affiliations of the radios to the base stations and assign the traffic channels of those base stations. 

Early TETRA networks were based on circuit switched technology using time-division multiplexing (TDM) and synchronous protocols such as E1 and T1 for the links between the switches and base stations. However, E1 networks require a complex architecture and configuration, with manufacturers providing costly proprietary hardware. Energy, implementation and maintenance costs are high.

“Components for traditional TDM-based circuit switched TETRA networks are becoming obsolete now,” says Bert Bouwers, CEO of Rohill. “These networks are also less flexible, which makes it harder to implement new features. Yes, you can do E1 over IP, but it causes problems. You can easily get network disturbances with the slightest delays or dropped IP packets, and that means bad performance. Today, networks require full IP links to the base station.” 

Manufacturers started to use IP-based elements for TETRA networks in the early 2000s. The introduction of packet switching of voice, data and signalling did provide significant improvements compared with circuit switching, including a simpler architecture with fewer hardware components and more flexibility and integration opportunities. But proprietary elements were still required to optimise IP for TETRA.

The next step was to move to full IP networks using Ethernet connectivity based on open standard commercial off-the-shelf (COTS) IP routers and commercially available software, thereby avoiding the need for proprietary hardware. 

As of 2017, the main TETRA infrastructure manufacturers provide full IP-based solutions, albeit with different approaches to network architectures.  

Rohill adopted what it refers to as the soft switch-based technology approach in 2002. Standard IP software deployed on top of COTS server hardware is used to perform all processing, switching and routing of TETRA speech and data with IP links to interconnect the systems and base stations. “We were very strict about only using open standard IP-based protocols with no proprietary elements,” says Bouwers. 

“There are a number of advantages to using the soft switch approach. We investigated what kind of networks work well with IP/multiprotocol label switching (MPLS). MPLS is the kind of core IP network technology that is required to make a robust mission-critical TETRA network. If you use commercial-grade IP links or go over the internet, you have no control over the links and that is not reliable,” argues Bouwers. “But if you use the IP/MPLS approach it is easy to achieve redundancy within the network [by duplicating the soft switch] between the systems and the base stations. If a soft switch goes down, services will switch over to another soft switch in under 50ms with almost no packet loss.”

Bouwers believes this is highly beneficial for planning and running a network, because there are no longer complicated dedicated E1 links to maintain. He says the unique point of the Rohill solution is that it only needs one Linux server to run the whole TETRA system. “One Rohill TNX node can support up to 120 base station sites using just one server, so even a big city only needs one TetraNode Core system.”

Rohill has adopted a hierarchical switching approach with switches at city or regional level. The logic for this is that it keeps all the traffic local. “We’ve found that 80-90 per cent of all communications are within a city,” says Bouwers. “You can then connect different cities or regions together at a higher level using multi-node connections. We have just migrated three TETRA networks in Poland for Warsaw, Krakow and Szczecin. Each city has a separate node, but they are all configured together to become a single TETRA network.”

Motorola Solutions 
Motorola Solutions also made an early move into the IP-based switching world, back in 2001. Its first virtualised platform was released in 2011 in the shape of the Dimetra Release 8.0, also referred to as Generation 3 hardware or infrastructure. “That put us on the road of a more COTS-equipment-based environment, using things like HP standard IP servers,” says Kevin Humphrey, business development director at Motorola Solutions’ TETRA Solutions Group. “Whereas before different applications used to be hosted on individual servers, we could host them on three core servers.

“Using COTS equipment makes it cheaper for us as we do not need to maintain legacy equipment. Instead, the COTS equipment manufacturers keep maintaining and developing the equipment to keep it up to date.”

The company’s latest generation switch was unveiled in May at CCW Hong Kong, when the Dimetra X Core solution was launched, along with the Dimetra 9.0 software. This provides a completely virtualised platform on a single server, rather than the previous three, capable of handling any network from one site up to 5,000 sites or more.

“X Core has a software-defined core,” explains Humphrey. “The hardware is in virtualised containers with the applications in store. If a customer wants to add an additional feature they just need to add a software licence and we can just turn it on. Previously, we’d have had to introduce the new application and configure it within the system.

“It is very easy to add new features because everything is based on open standard IP systems such as Linux and Windows hosted on open standard virtualisation software. None of that is proprietary and that means we can focus on what we do best, which is providing TETRA applications.”

From the physical point of view, the virtualisation behind the Dimetra X Core reduces the amount of hardware required, providing a smaller footprint, lower power consumption and reduced maintenance costs.  

Airbus also moved further into the IP world in 2017 with the launch of its Taira TETRA Server, which is the sequel of the DXT switch series. Mika J. Laitinen, head of TETRA portfolio management of Secure Land Communications at Airbus, explains.

“Airbus has been closely working together with its customers, and was also deeply involved in ETSI and the 3GPP standardisation process. This supports end users’ current and future operational models and ensures that their TETRA technology can operate in the future. Our conclusion is that the TETRA networks and broadband services legacy will still co-exist for a long time”, he continues. 

 “But to deal with this need for the co-existence of TETRA and LTE we had to develop a product strategy which offers synergies with the technology platforms that are also used for 4G and 5G. And we worked out what that means for legacy TETRA technology. 

“This is the background to what is happening with the Taira Server, where we are now changing from proprietary hardware to COTS servers using the same virtualisation environments that our customers use for 3GPP networks. 

“But even if it is a new generation TETRA product, end-users will not see any change in their daily operations. They get exactly the same services and functionalities as they are used to having with our previous generation TETRA system,” says Laitinen. 

The Taira server also makes use of commercial virtualisation software. “If you look at the TETRA core today it has multiple elements, gateways, databases and so on. When we virtualise TETRA core elements into the same server cluster, our customers have both Capex and Opex savings. By virtualising system aspects, we can enable all the core elements to run on one server cluster. This streamlines network operations for our customers and reduces the amount of hardware needed with less footprint,” says Laitinen.

He adds that, as the Taira server is a fully IP-based solution, it is easy to install a TETRA network in a customer’s existing data centre if they wish to do so. “We are integrating more functions into the core than was possible before. It is easy to bring in other IP-based systems and you no longer need the specialised skills to operate and maintain proprietary TETRA hardware, so from an equipment maintenance point of view it is much easier. The new solution allows customers easier use and flexibility,” he says. 

Hytera opted for a full open standard IP approach around 2008-09 for its TETRA ACCESSNET-T IP system architecture, according to Thomas Conrath, specialist product manager for TETRA infrastructure at Hytera Mobilfunk. Hytera’s system supports both centralised and fully distributed architectures, or even a mix of the two.

“We now design and deliver a system that is full standard IP with standard COTS equipment such as routers and servers, commercial firewalls and operating systems, so there is nothing proprietary in the system as far as the protocols and equipment are concerned,” Conrath says. “We use a flat architecture, even if we are talking about multiple regions.”

An all-IP architecture makes network architecture and configuration much simpler. “If you no longer need to define all the E1 links and all the slots and sub-slots between the base stations, configuration becomes really easy,” says Conrath. “You can use existing IP structures and multiple applications. There are lower training costs and the customer does not have to hold high stocks of proprietary spares.”

However, Conrath warns that operators need to be aware of the product lifecycle of commercial equipment. If the likes of Dell, Cisco, IBM and HP decide on a specific lifecycle, and for modern server equipment it is about three years, customers need to keep that in mind. 

He notes that virtualisation is playing a bigger role, helping to improve energy efficiency and making it much easier and cheaper to realise redundancy and efficiency enhancements. This leads to reduced downtime in daily operations and therefore a huge increase in terms of network availability. “A modern TETRA network is essentially software and hardware base stations providing the air interface. It is almost just base stations plus a CD with the software,” says Conrath. 

Damm offers a fully IP-based platform based on a flat distributed network architecture, which includes base stations, dispatchers, network management tools, external gateways and other applications. Its TetraFlex system can scale up to handle thousands of sites without any centralised switch.

Jochen Bösch, head of support and product management at Damm, says: “For us, the use of pure Ethernet connections started in 2005-06. TetraFlex was designed for use straight out of the box. It is a fairly flat architecture, so the customer is free to choose the topology, be it star or ring. Any IP backbone can be used, be it DSL, leased lines, fibre, LTE or satellite.” 

The big difference with Damm is its use of a fully distributed architecture. This involves distributing the controller intelligence normally contained in a centralised switch or several switches to the base stations at the edge of the network. 

Damm believes this approach provides a highly reliable network, as the intelligent distributed network architecture means all system information is constantly replicated to all sites in the network. In this way, local call and data traffic will always continue uninterrupted, as each base station contains the TETRA feature set normally housed in a centralised switch. 

“A decentralised architecture is much more robust against delays and jitter effects in the backbone,” explains Bösch. “Our recommendation is to have gateways at each base station, as this way the network is more reliable and will not be hit by the failure of a centralised system, relying on one or two switches.” 

Most other manufacturers use a centralised switching system, at least for larger networks, although they are comfortable providing fully distributed systems for small to medium-sized networks of up to 25-30 base stations. 

The proponents of centralised TETRA switching argue that it becomes increasingly difficult to scale up a fully distributed system, as each base station needs to constantly keep all the others informed as to what is going on within the network, making it tricky to predict the bandwidth requirement and handle the wealth of database synchronisation messages. 

Bösch professes himself bemused by these arguments: “We can scale up the network to an unlimited number of base stations because we take a good approach in terms of subscriber management. We reduce the potential traffic between nodes by just sending the updates, so you don’t get a high dataload.”

He argues that the bandwidth requirement and database synchronisation are both fully plannable and fully configurable – and will remain unchanged if the network is scaled up. In fact, he says they are equal or less than in a centralised set-up. 

Network migration
The arguments over the merits of centralised or distributed switching will no doubt continue, but whichever architecture is chosen, migrating live networks to all IP systems needs to be carefully handled to avoid any disruption.

Motorola’s Humphrey says: “We support a slow migration approach to the introduction of new infrastructure. What you can do with an existing Dimetra system is connect with Dimetra X Core and slowly migrate to it, and that reduces the downtime and impact on users.”

Hytera’s Conrath adds: “Our software can handle both IP and TDM E1 operations at the same time. It’s a parallel operation of switching in IP and TDM E1 in one common network with a common administration. The customer can then decide whether to adopt a single-step migration path or just switch over one base station per day.” 

Security issues
The use of mass-market hardware and software naturally raises concerns over the security of modern IP-based TETRA networks. “Using COTS hardware and software does bring some challenges for our customers,” acknowledges Humphrey.

“What you have now is basically an IT network with the base stations providing the radio frequency interface to the users. You need to ensure you have anti-virus software and be able to patch the system if required. That requires a shift in thinking for our customers,” he says. 

The upside of using COTS solutions, observes Conrath, is that somebody else has usually come up with best practice guidance for security, along with commercial firewalls and anti-virus software. “You can pretty much rely on worldwide resources to come up with fast patches to deal with malware,” he says.

It is clear that TETRA networks based on open standard IP switching running on COTS hardware and software provide many benefits for customers. As Rohill’s Bouwers observes: “What customers really appreciate is the ease of configuration and management.”

A further advantage of moving to an all-IP-based TETRA switching solution is that it makes it easy to integrate with LTE broadband networks, widely seen as the future for critical communications, although most believe TETRA and LTE will co-exist for a long time to come.

Author: Tetra Today