Ultra-wideband is already capable of remarkably precise indoor positioning. The bigger challenge is turning that precision into an industrial-scale system that is robust, scalable, and economical to deploy. Ciholas President & Founder Mike Ciholas explains how the company’s CUWB platform approaches that problem – including a warehouse deployment delivering 7,000 location calculations per second.
Ultra-wideband (UWB) is increasingly associated with precise indoor location, but achieving centimetre-level accuracy is only part of the challenge. For industrial deployments, location systems also need to operate reliably across very large facilities, cope with obstructions and changing RF conditions, support thousands of moving assets, and do so at an infrastructure cost that delivers an acceptable return on investment.
That is the problem Ciholas Inc. has set out to solve with its CUWB (pronounced ‘cube’) real-time location system. In a recent WiFi NOW TECH WATCH interview, Ciholas President & Founder Mike Ciholas presented the company’s approach to what he calls “enterprise-grade indoor location.” At its simplest, the architecture resembles a private indoor GPS system. Tags are attached to people, machines, or materials while fixed UWB anchors are distributed throughout the facility. The anchor network feeds a backend location engine, which calculates the tag’s identity, time, three-dimensional position, and location quality before delivering the information to applications for tracking, analytics, safety, inventory management, or security.
Industrial UWB at warehouse scale
Perhaps the most striking example presented by Ciholas is a CUWB deployment in a 130,000-square-metre (1.3 million square foot) fulfilment warehouse. The system tracks 300 forklifts at 10 Hz and 2,000 personnel tags at 2 Hz, resulting in 7,000 location calculations per second. Ciholas says the installed system has capacity for up to 10,000 locations per second. The warehouse extends approximately 700 metres by 185 metres.
This scale matters because industrial location is fundamentally different from demonstrating high positioning accuracy within a laboratory or a single room. A warehouse contains shelving, machinery, people, vehicles, and other objects that can obstruct or distort radio signals – while the system still needs to determine where a moving asset is in real time. Ciholas addresses this partly through its location engine. Rather than relying on a fixed number of anchors for each calculation, the engine can use large numbers of available anchors and assign different weights according to measured signal quality.
In one example presented during the interview, a forklift moving through an open travel way can be located using signals from many anchors – sometimes as many as 100. When the forklift enters an aisle between warehouse racks, anchors affected by obstruction or interference are given less weight, while anchors delivering higher-quality measurements receive greater weight in the location solution. Ciholas describes the approach as a best-fit location algorithm, designed to find the position that best matches the measurements produced by the anchor network.
From centimetres to millimetres
The accuracy figures presented by Ciholas are noteworthy – and exceptional. In laboratory testing without the obstructions and racking of a real warehouse, six anchors produced standard deviations of approximately 2.4 cm in X and 2.3 cm in Y. With 12 anchors, those figures fell to 1.5 cm and 1.4 cm respectively, while 24 anchors produced approximately 0.8 cm in both dimensions.
The system can increase precision further by taking multiple measurements and averaging them. In the example presented by Ciholas, a tag operating at 100 Hz was averaged into a single position output each second. With 24 anchors and 100-sample smoothing, the presentation reports standard deviations of 0.2 cm and 0.3 cm – or two and three millimetres respectively.
That extreme configuration is not necessarily representative of a typical industrial deployment. Ciholas instead describes accuracy as an engineering and economic trade-off: A customer can deploy fewer anchors where cost is the priority, or increase anchor density and smoothing where extreme precision is required. The laboratory measurements should therefore be understood as demonstrations of the system’s achievable precision under controlled conditions, rather than universal accuracy figures for every deployment.
A single synchronised UWB network across the warehouse
Scaling UWB across a building introduces another engineering challenge: Synchronisation. The warehouse deployment uses approximately 650 anchors, which Ciholas says are synchronised into one contiguous timing domain. The objective is to avoid dividing the installation into independent location clusters or zones, where transitions between one group of anchors and another could introduce anomalies as a tracked object moves through the facility.
This is an important distinction between UWB as a short-range positioning technology and UWB as an industrial infrastructure platform: Large deployments require not only accurate individual measurements but also consistent positioning behaviour across the entire operational area.
UWB doubles as data collection (connectivity)
CUWB also uses the UWB radio link for more than positioning. The system supports bidirectional data communication between a tracked device and the backend. A forklift, for example, can send information including inertial measurement data, speed, direction, pose, operator identity, or SKU information.
Commands can also travel in the opposite direction, including speed limits and zone-based stop commands. According to Ciholas, this can remove the requirement for a separate Wi-Fi or Bluetooth connection within the tracking infrastructure for these functions.
That does not make UWB a general replacement for Wi-Fi. Rather, it allows the location system itself to carry relatively modest amounts of operational data associated with the tracked asset.
Cutting infrastructure costs with chained PoE
High positioning accuracy has limited commercial value if deploying the anchor infrastructure becomes prohibitively expensive. Ciholas therefore puts considerable emphasis on total cost of ownership. One notable design is the company’s chained Power over Ethernet architecture.
Each anchor has two Ethernet ports, allowing as many as 12 anchors to share a single PoE+ connection. Ciholas says this architecture can reduce the number of switch ports and wiring required by approximately 90% compared with individually wiring every anchor. The design can also be configured redundantly by connecting the end of the anchor chain back to the network.
Ciholas says this means an individual cable failure does not take the entire chain of anchors offline. Power requirements are low as well. Ciholas says each anchor consumes less than two watts, making it possible to operate a 12-anchor chain from a 30W PoE port The company’s broader TCO proposition combines relatively low equipment and infrastructure costs, reduced installation effort, and no recurring software or licence fees, according to Ciholas.
UWB and Wi-Fi: Side-by-side rather than inside the same box?
An interesting part of the discussion concerns whether UWB infrastructure will eventually become integrated into enterprise Wi-Fi access points. Ciholas is sceptical – particularly for large-scale location deployments – because he sees fundamental differences between the optimum placement and infrastructure requirements of the two technologies.
UWB anchors may need to be distributed around the perimeter and throughout a facility to achieve suitable positioning geometry. Wi-Fi APs, by contrast, are generally positioned to provide efficient RF coverage and capacity. Their infrastructure requirements also differ: Modern enterprise APs typically require high-speed Ethernet and substantially more power, while Ciholas says its UWB anchors require less than two watts and can operate with 100 Mbps Ethernet.
His conclusion is therefore that Wi-Fi and UWB infrastructure can coexist in the same building but will not necessarily benefit from being integrated into the same physical device. That is one vendor’s architectural view rather than an industry-wide conclusion. Other vendors are pursuing combinations of Wi-Fi and location technologies within enterprise infrastructure, so how these architectures evolve remains an open question.
Industrial tracking versus indoor navigation
Ciholas also makes an important distinction between industrial tracking and the emerging concept of untracked indoor navigation. CUWB’s industrial model centrally determines the location of tagged people or assets. It is therefore particularly suited to controlled environments such as factories, warehouses, and other facilities where the organisation operating the infrastructure is also responsible for the machines or personnel being tracked. Untracked indoor navigation reverses that model.
Infrastructure broadcasts timing information, while a user’s device calculates its own position without the infrastructure knowing who is receiving the signals. Ciholas compares the concept to bringing the GPS experience indoors – allowing a phone to continue determining its own location inside airports, hospitals, railway stations, retail environments, and other public spaces.
The two models could ultimately address very different parts of the indoor-location opportunity.
Is industrial UWB nearing an inflection point?
Ciholas believes the combination of higher performance and lower deployment cost could expand UWB beyond today’s relatively specialised installations. Warehouses and factories are obvious candidates. Accurate knowledge of where people, vehicles, tools, materials, and inventory are located can feed applications ranging from logistics optimisation and automation to worker safety and asset management.
Ciholas draws an analogy with the evolution of Wi-Fi in warehouses: Twenty years ago ubiquitous warehouse Wi-Fi was far from guaranteed, while today wireless data connectivity is fundamental to warehouse operations. His argument is that UWB location infrastructure could follow a similar path if its performance improves and its deployment economics become sufficiently attractive.
Whether UWB ultimately becomes as ubiquitous in industrial environments as Wi-Fi remains to be seen. But the CUWB deployment presented here demonstrates that the technology is already capable of operating at substantial scale: thousands of tracked assets, thousands of location calculations per second, hundreds of synchronized anchors, and – under suitable configurations – positioning precision extending from centimetres into millimetres. For industrial UWB, the next challenge may therefore be less about proving that precise location works – and more about making that precision sufficiently robust, scalable, and cost-efficient to become everyday infrastructure.
/Claus.