Maximizing Data Analytics and AI with Global IoT Connectivity

Real-time decision-making
using IoT telemetry data

Company’s global assets – their products, industrial machinery, or fleets of vehicles – are increasingly connected, capturing detailed user and operational data.

Combined with cloud and analytics, this data can deliver valuable business insights – such as predicting machinery faults, spotting when someone needs healthcare interventions, or real-time planning of logistical operations.

But models are only as good as the data going in. Garbage In, Garbage Out, as the computer science adage goes. If we are building clever models on data from fleets of thousands of geographically dispersed devices, we need to be sure that they are sending that data back as expected. This needs reliable and highly available connectivity.

On this page we look at the value of IoT data for business, the challenges of getting reliable data from dispersed IoT devices, and how we can design IoT products in ways that overcome these challenges.

Fleet management goes beyond vehicles, combining telematics data on the environment in which they operate."

Smart asset performance management

Many machinery manufacturers are implementing smart asset performance management, connecting their machines, and using data to predict repairs, maintenance, and refills. A South African spice company, Freddy Hirsch, makes food ingredients and industrial food machines. Rather than selling products separately, it now sells its IoT-enabled sausage machines.

Enabled by cellular connectivity, Freddy Hirsch can detect when ingredients are low and resend when they are needed. It can spot problems by looking for key signatures of impending faults in the stream of usage data coming off the machine, and pre-emptively send engineers to fix them.

Costa Coffee has a growing line of business through vending machines. It collects streams of data on usage, boiler temperature, supply levels, etc. This allows it to spot where it needs to send engineers and additional supplies. And as it acquires more data, it can build models that support pre-emptive maintenance and supply chain optimisation.

In both cases, the regular flows of telemetry data coming off the machines are vast and unlikely to be manually analysed. If this data is incomplete due to machines dropping connectivity, it can lead to missed opportunities for maintenance and supply chain efficiency.

Smart Healthcare

Telehealth is allowing older people to live comfortably at home. Alcuris’s Memohub connects smart devices around a person’s home, such as kettles and televisions, and backhauls this data to the Alcuris AWS cloud database, where it performs analytics. For example, it can establish patterns of behaviour – such as the time the kettle is turned on each day and monitor gradual changes which may indicate a need to review care regimes or even alert to a potentially life-threatening situation.

This needs reliable data over time to build models of usual behaviour, so that it can detect deviations of concern and alert carers if care regimes need to be changed. If the ‘kettle turned on’ data point is several minutes late every other day due to connectivity dropping, rather than actual behavioural change, the model won’t be looking for the right thing, and may miss – or misdiagnose – real problems when they count.

For more information, see how millions of vulnerable people to benefit from new IoT monitoring service.

Ever smarter opportunities

As our data capabilities become more sophisticated, this IoT data will be used to do much more. Telehealth data may help drug developers study the impact of medicines which may save lives; industrial machinery data may help energy providers finely balance energy production to reduce emissions.

These models will rely on aggregates of multiple vast datasets to find key predictive signals. Patchy data not good enough. As IoT data feeds ever more sophisticated models, it is important that the data being collected is as reliable, consistent, and accurate as possible.

Even if current models can deal with a few gaps, establishing good IoT data collection now will set you up for the data-driven future.

What causes the problems
for IoT data collection

Models can be hindered by data that is poorly labelled, hard to find, or incomplete. Some of this needs to be addressed by good data collection processes that go beyond this paper. We will look specifically at the potential for poor connectivity to deliver incomplete or inconsistent IoT data collection. This comes in two flavours.

Geographical

This refers to when some devices are – permanently or temporarily – in an area with no connection and are unable to send data back. This can happen when a device is positioned out of the network range (as with some road measurement devices) or moves in and out of it (as with vehicles).

Any phone user knows that mobile networks have reception black spots. Deploying thousands of assets across a country on a single mobile network will mean 15-25% without a reliable connection. This will lead to a significant number of data blind spots.

Given the nature of IoT, devices are often highly mobile and therefore run the risk of being disconnected from the network when they move. If a network has high latency the issue is exacerbated further and impacts, too, on bandwidth. IoT devices rely on a continuous connection to a responsive network, otherwise packets of data are delayed. For every millisecond of delay, this impacts IoT device’s ability to work at its full potential and transmit accurate, real-time data.

Temporal

This is where the stream of data being fed back is interrupted due to networks being temporarily unavailable and connectivity being lost. Some devices need to provide a constant stream of real-time data. Temporal connectivity issues can interrupt monitoring and have serious repercussions.

Several factors can lead to connectivity loss. Fragmented connectivity coverage. Networks sometimes break. Roaming agreements change. Regulations take force. Consumer-grade SIM cards deliberately drop connections to idle devices, which can take time to reconnect, delaying transmission of time sensitive information.

If the device is the driver of IoT project success, connectivity is the backbone; everything will depend on it."

Designing an IoT solution for
reliable data collection

In this final section, we look at how to design in connectivity to overcome these problems and ensure a reliable stream of data wherever your connected device or asset is in the world.

Device design

The first step of ensuring reliable connectivity – and so reliable streams of real-time data – is to minimize how often it is unable to transmit data. But nothing is 100% perfect, so the second step is to ensure the system behaves well if it goes wrong. Even for those skilled in hardware and software, designing compliant connected devices is a whole new challenge. We recommend an onboarding test that simulates the most disruptive network conditions possible, to see whether the system still functions acceptably and recovers well. This will help identify not just challenges for the SIM, but how to design and set up hardware and software to deal with problems. A simple example would be building in backups so that data points are time-stamped and transmitted in the right sequence the moment it reconnects, and the cloud platform is set up to correctly interpret delayed data.

Optimizing the battery is another aspect to be considered during the device design process. It’s important to ensure that battery-powered IoT edge devices aren’t awake all the time and only transfer data when they need to. Energy can quickly be expended if devices are sending regular packets of data – no matter how small. It’s worth looking into LPWA key power reduction features such as eDRX (extended Discontinuous Reception) and PSM (Power Saving Mode) to improve battery efficiency and futureproof the device for when it’s out in the field.

LPWA Quick Definition: LPWA (Low Power Wide Area) is a type of wireless communication technology designed for IoT (Internet of Things) devices that require low power consumption and long-range connectivity. LPWA networks, like NB-IoT, LTE-M, and LoRa, are ideal for devices that transmit small amounts of data over extended distances, often in remote or hard-to-reach locations, while minimizing battery usage for prolonged device life.

Use a multi-network eSIM that maximises geographic availability through localization

Use a single SIM technology (such as Eseye’s AnyNet+ eSIM) capable of switching remotely between multiple networks. This will enable the device to find the best connectivity no matter where it is in the world, and switch easily between networks when it’s on the move.

SIMs need to load an ‘IMSI profile’ – to authenticate themselves for each mobile network. Many SIMs only allow one profile, so are limited to one network and its roaming partners. In contrast to standard eSIM solutions, which offers only a single bootstrap fallback option, Eseye’s AnyNet+ SIM combines GSMA-compliant eSIM technology alongside the capability to use up to 10 bootstraps installed on the SIM and the ability to load a new IMSI over-the-air (OTA). This affords it total flexibility and means it can move between a wide range of networks, giving much wider coverage and reducing lost revenue which might occur because of a network service outage. Essential if you want reliable data transmission regardless of location.

The second part of this is ensuring your connectivity provider has agreements with all the global mobile network operators whose networks your device may need to call on. There’s no point having a multi-IMSI SIM if it’s not actually allowed to join half the world’s networks.

Note that roaming SIMs are not a long-term nor reliable solution. Many organizations run into trouble when using them with significant cost implications to ensure their devices remain connected. Some countries, such as Turkey and Brazil, impose restrictions on devices roaming for more than a few months, and tighter permanent roaming rules look to be coming in many other areas. Agreements with local operators, not just the big global ones, is also important for a true global deployment.

Ensure reliable cellular connectivity

Making the most of your SIM’s ability to switch networks, and so transmit data to the cloud, requires a connectivity management platform (CMP). The CMP monitors device connection and data flow and instructs them when to update their profile to access the best network, when the device moves or the network falters.

A global SIM should be set up to reconnect to the best network if it loses reception. But with a CMP, it can constantly monitor for any drop in quality of data transmission and move it onto a better network. Active management allows real-time optimization of all devices across your network. It moves you from ‘having connectivity’ to ‘always having the best connectivity’. This is very valuable if you are reliant on real-time data.

CMP Quick Definition: A Connectivity Management Platform (CMP) is a centralized tool for managing and optimizing IoT device connectivity across multiple networks and regions. CMPs enable businesses to oversee and control SIM activation, network selection, and data usage while ensuring reliable device connectivity and simplified operational management.

Eseye’s Infinity IoT Platform™ goes beyond typical CMPs by providing complete control over your IoT connectivity. With Infinity, you can manage all aspects of IoT connectivity—from SIM provisioning to seamless global network switching—with eUICC compliance for maximum uptime.

Cloud & application integration

Usually, you will want your data to feed into a cloud platform or enterprise application where you can analyze data in detail using AI and analytics models. This ‘backend’ will then process the data and present the insights in easy-to-use dashboards, allowing decisions to be made quickly. For example, both AWS and Microsoft Azure’s cloud platforms provide specialist IoT data analytics capabilities.

The route to integrating IoT data with these cloud platforms lies in the deployment of the application programming interface or API. This is a standard protocol for exchanging two-way information between systems. Connectivity, enterprise applications, and cloud providers may offer a range of purpose-built APIs designed for sharing different types of IoT data. Talk to them about your data sharing needs.

Data provisioning and security

Security is vitally important when data is transmitted and provisioned from the device over-the-air to cloud services. It’s critical to use security certificates to identify the device to the cloud and to send encrypted data over secure communications channels. Some IoT SIMs are basic and overlook these important security issues.

Agentless device security platforms exist to address the new threat landscape of unmanaged and cellular IoT devices. For example, together the leading agentless IoT security provider Armis and Eseye now provide a unified security and IoT connectivity solution across all IoT devices, including those on local business networks (Ethernet, Wi-Fi, etc.) and mobile (4G/LTE/5G) networks.

Try Eseye’s Free IoT SIM Trial and Device Assessment Kit with a prepaid data plan and AnyNet+ trial SIM—plus expert testing and insights to optimize your IoT deployment. Claim now →

Conclusion

Data from IoT devices is becoming ever more valuable. Complex processes – from health alerts to route planning – involve taking in telemetry data from IoT sources around the world on a scale beyond human comprehension, running it through clever models, and reaching precise recommendations.

Many offer incremental efficiency improvements that add up to big savings across vast networks of assets and products. Others deliver critical warnings that may avoid serious adverse events.

Ensuring this data arrives securely, as expected, at the right time, and in the right order is critical to getting the most out of these systems and benefiting from the transformational insights that rich data can offer.

At the heart of reliable global IoT data collection is reliable global connectivity.