Deterministic industrial wireless in practice: the AIMesh™ 6TiSCH network infrastructure

01 The determinism problem in industrial wireless

Digital transformation in Industry 4.0 and smart manufacturing rests on connecting every production element. Wired networks — industrial Ethernet and fieldbus — offer stability and strong real-time behaviour, but across large steel-framed buildings, complex oil and gas stations, tank farms and underground facilities, wired deployment runs into high cabling costs, difficult maintenance and a lack of flexibility.

Wireless is therefore an indispensable extension of industrial IoT. But conventional commercial wireless — ordinary Wi-Fi, ordinary ZigBee — uses non-deterministic contention-based channel access, which makes it vulnerable to multipath fading, industrial electromagnetic noise and co-channel interference, producing packet loss and severe latency jitter. For industrial control and monitoring that need high reliability and bounded latency, that is a decisive shortcoming.

To break through this, AISENZ developed AIMesh™, a next-generation industrial wireless IoT protocol and product family. As a wireless infrastructure platform for the next generation of industrial IoT, AIMesh™ deeply optimises the 6TiSCH (IPv6 over TSCH) architecture, combining novel radio modulation, distributed scheduling algorithms and edge AI into a deterministic industrial wireless foundation that is low-latency, highly reliable and ultra-low-power.

The hard part of industrial wireless is not "can it connect". It is whether it can keep delivering predictable latency, reliability and lifecycle cost across a difficult site.

02 Core architecture and underlying mechanisms

What distinguishes AIMesh™ is how it introduces determinism into an inherently non-deterministic industrial radio channel. Its foundation is the IETF 6TiSCH stack, optimised end to end for the industrial field.

TSCH time-slotted channel hopping

TSCH (Time-Slotted Channel Hopping) is the core mechanism for deterministic wireless. It divides time into small slots and the available spectrum into multiple orthogonal channels.

• Slot synchronisation: every node on the network holds microsecond-level time synchronisation. Which slot a node transmits or receives in is set by an explicit schedule, eliminating the collisions that arise when nodes contend for the channel.
• Channel hopping: each transmission moves the working channel according to a predefined pseudo-random sequence. Even if one band suffers severe industrial electromagnetic interference at a given moment, the next slot hops to a clean channel, improving resistance to interference and fading.

Headline performance

By optimising the 6TiSCH stack and strengthening distributed scheduling, AIMesh™ achieves engineering-grade performance on live industrial networks:

• End-to-end reliability: multipath routing plus time and frequency diversity give an end-to-end delivery rate of approximately 99.999%, meeting the process industry requirement that critical monitoring data be lossless and traceable.
• End-to-end latency: optimised scheduling compresses queue waiting time, holding average end-to-end latency under 1 second — sufficient for low-latency industrial interlocking and wireless closed-loop systems.
• Ultra-low power: precise slot-based sleep control keeps the node radio off when idle, so low-power nodes reach five to ten years of battery life depending on reporting frequency and routine field instrument maintenance costs fall.
• Native IPv6: AIMesh™ carries IPv6 across the whole network, giving each field node a unique IPv6 address so data reaches the enterprise backbone, SCADA systems or cloud platforms end to end.

TSCH moves the radio link from probabilistic contention to deterministic scheduling; IPv6 turns field nodes from isolated devices into network assets that IT and OT systems can address uniformly.

03 Topology and core components

AIMesh™ uses a distributed, modular component design in which different node functions cooperate to form a self-organising, self-healing mesh.

Leaf and routing nodes

• Leaf nodes: mostly battery-powered low-power field instruments with an embedded communication module. They sit at the network edge in deep sleep, waking only for scheduled acquisition or a burst alarm, and carry no forwarding duty — which is what buys the longer runtime.
• Routing nodes: typically continuously powered field devices or dedicated repeater modules. Beyond their own acquisition, they hold the backbone topology, maintain multipath routing tables dynamically and provide multi-hop forwarding for distant leaf nodes.

AP01 border router gateway

AP01 is the bridge between the wireless network and the wired backbone — an IPv6 border router. Downstream it joins the AIMesh™ wireless network, managing node admission and network clock synchronisation; upstream it aggregates IPv6 packets over industrial Ethernet. Across a large plant, several AP01 units deploy together to extend coverage, raise total capacity and provide redundant access across zones.

E680 edge computing and control gateway

E680 sits at the compute layer of the architecture, handling local industrial protocol conversion, real-time local processing and buffering, and local closed-loop control. It is also the physical host for the AIMesh Manager network management software, providing compute for the whole wireless network.

Field nodes handle sensing and access, AP01 handles the wireless-to-wired border, and E680 handles edge compute, protocol convergence and closing the business loop.

04 The product family and key specifications

To span lightweight sensor embedding, heavy industrial equipment connectivity and high-compute edge control, AIMesh™ offers a complete product matrix.

AIMesh™ M01 wireless module

M01 is an embedded wireless module for industrial instrument manufacturers, letting a conventional field instrument go wireless with a short development cycle.

• Core and interfaces: an ARM Cortex-M4 with integrated FPU keeps local protocol parsing and hopping schedules real-time. UART1 carries standard AT commands, UART2 carries Modbus RTU or a custom binary stream, connecting to mainstream instrument MCUs.
• Radio characteristics: operating in the 2.4 GHz ISM band with −106 dBm receive sensitivity and 12.5 dBm maximum transmit power, giving useful penetration and range within a micropower budget.
• Engineering convenience: SMT reflow mounting, a standard IPEX antenna connector, and on-site commissioning over Bluetooth from a phone app, which lowers the effort of bringing a network up.

AIMesh™ D01 industrial DTU module

D01 is an external industrial data transfer unit, used to digitise existing wired equipment or to act as a standalone repeater.

• Seamless retrofit: D01 carries one standard RS485 interface supporting Modbus RTU or fully transparent transmission. An existing wired instrument only needs its 485 bus connected to D01 to gain AIMesh™ mesh network access.
• Topology optimisation and standalone relay: in dead spots where steel structures block the signal badly, D01 can be installed at a key location as a pure relay routing node with no instrument attached, extending coverage and improving the overall topology.
• Industrial enclosure and mounting: ABS enclosure with standard DIN rail mounting. The configuration port is USB Type-B, used for AT command access and temporary power.

AIMesh™ AP01 border router gateway

AP01 is the aggregation point between the wireless mesh and the enterprise industrial Ethernet backbone.

• Processor and ports: a 3× Cortex-A7 @ 1.5 GHz multicore processor handles packet processing and forwarding for hundreds to thousands of concurrent field nodes. Interfaces cover one AIMesh radio, one RS485, two Ethernet ports and one USB.
• Radio enhancement: compared with the end-node modules, AP01 strengthens the RF front end — maximum transmit power rises to +19 dBm and receive sensitivity to −111 dBm — so it can hear weak signals from distant instruments and keep the backbone link stable.

E680 edge computing and control gateway

E680 is the compute flagship of the family, acting both as manager of the wireless network and as the point where edge AI reaches the industrial floor.

• Hardware: an 8-core processor of 4× Cortex-A72 plus 4× Cortex-A53, with a 6 TOPS @ INT8 NPU and an ARM Mali-G52 MC3 GPU. Interfaces cover Gigabit Ethernet, USB Type-A, RS485, CAN-FD, GPS and a Wi-Fi expansion module.
• Edge AI and LLM inference: local compute lets E680 run lightweight large language models such as Qwen 1.5B / 3B on site, for self-explaining operating conditions, fault root cause analysis and control strategy suggestions, and for natural language interaction with operations staff.

From M01 and D01 through AP01 and E680, AIMesh™ is not a single device but an infrastructure set spanning field access to edge intelligence.

05 Field deployment and parameter tuning

In real deployments there is a standing tension between a complex electromagnetic environment and device power budgets. AIMesh™ provides software components and configuration options for balancing performance, coverage, power and operating cost.

Core deployment parameters

When deploying AIMesh™ nodes — the M01 module here — engineers tune the following against each device's characteristics:

Preloaded software and visual management

The E680 edge gateway supports a broad industrial software stack, preloadable per project: edge AI inference modules, a Python runtime, the PatchTST time-series forecasting model, lightweight LLMs, SoftPLC logic control, plus SQLite, Nginx, Node-RED and MQTT middleware.

In day-to-day operations, the AIMesh Manager running on E680 gives a transparent, visual view of the network topology for remote operations. Users log in from a standard browser and watch the topology evolve in real time. Link quality across the network is colour-coded by packet delivery ratio: green (PDR above 99%, good), yellow (95–99%, moderate) and orange (below 95%, poor).

The goal of parameter tuning is not the best-looking number on a single node, but the engineering optimum for the whole network across reliability, latency, coverage and maintenance cost.

06 Typical industrial applications

Deterministic communication plus edge compute give AIMesh™ representative value across three classes of work: industrial monitoring, industrial control and industrial intelligence.

Monitoring: from process industry to infrastructure

In process industries such as petrochemicals, fine chemicals and metallurgy, AIMesh™ carries pressure, temperature and level monitoring across oil and gas production and storage, along with vibration and condition analysis on process equipment. On large bridges, dams, renewable energy plants, storage systems and data centre environmental monitoring, a reliable multi-hop mesh crosses geographic obstacles to provide long-term automated structural health monitoring.

Control: entering territory wireless was excluded from

With stable end-to-end latency under 1 second and distributed scheduling, AIMesh™ is progressively entering industrial control — territory conventionally closed to wireless. Applications include low-latency industrial interlocking, distributed joint control of local equipment and autonomous edge control, letting a plant build a high-integrity wireless closed-loop control system in areas where cabling is impractical.

Intelligence: predictive maintenance with AI at the edge

Combining E680's 6 TOPS of compute with built-in time-series forecasting, AIMesh™ performs high-frequency acquisition and edge AI inference on pumps, valves, motors and other rotating machinery locally, delivering anomaly detection, condition analysis and predictive maintenance — moving from repair after failure to prevention before it.