The natural transition from concept to IoT deployment starts with a Proof of Concept (PoC). PoCs are essentially pilot projects, there to evaluate the feasibility of a plan before it is implemented at scale.
Naturally, organisations want a means to determine if something can be deployed successfully before investing in a more mature IoT solution, however, some argue that focusing on an isolated IoT PoC may be the downfall of a successful deployment.
A study conducted by Cisco found that 60% of plans to implement IoT have been stalled at the PoC stage and only 26% of companies thought that their IoT initiatives was a complete success.
So, how do we make sure that PoCs are successful and can reach IoT maturity?
Three key areas, namely scalable architecture, artificial intelligence and security, may answer this question.
Scalable architecture for solution elasticity
Due to the nature of a PoC, the proposed concept isn’t intended to become an organisation-wide solution. Instead, scope is usually limited to prove certain outcomes.
The clear disadvantage is that architecture and solutions are designed and optimised for this scenario. Cost for example, may be prioritised over cohesion and flexibility, resulting in an assortment of components from different vendors, inhibiting control over components. Down the line, organisations may need to replace parts of their solution that were optimal for a PoC, but not suitable for a scalable and secure IoT capability.
The benefits of a broader scope with a long-term view can be illustrated when discussing the use of Virtual Private Cloud. This particular choice eliminates the need to create new servers each time the organisation grows, as it can store compute analytics modules and other microservices to the cloud.
Another consideration that may be dismissed in a PoC is the potential for the payload size to increase. Different data types can have significantly different payload sizes and bandwidth requirements. Something as simple as a water level sensor for example, only collects and transmits small amounts of data at a time. Contrasted with a device that has the ability to capture satellite images, the payload requirements are significantly higher.
Organisations need to take into consideration emerging IoT capabilities when they design their overall IoT architecture, to ensure that it is prepared to take on future challenges as the scope expands.
Artificial intelligence and deep learning
Gartner predicts that by 2023 one-third of companies with IoT deployments will also implement artificial intelligence (AI) in conjunction with IoT. AI and deep learning adoption is increasingly becoming a standard component of IoT solutions.
A mature IoT solution is able to take data collected from sensors and extract meaningful analytics and insights for an organisation to use. At the PoC level, however, a sensor might only collect and log data directly to a dashboard, without any further transformation.
This stands out as a major issue on its own. If a PoC is meant to prove the feasibility of an IoT initiative, then not utilising powerful aspects such as AI significantly limits the value it can provide to organisations.
Even if AI is adopted as part of the PoC, it may be unreasonable to expect instant results. Projects that require machine learning take a few months of concerted effort to collect and work through data to provide valuable insights. To expect this as an outcome of a PoC could be futile for the project.
Organisations need to consider the role that AI and deep learning play in the future and consider how to demonstrate its value to stakeholders.
AI and deep learning insights from camera footage
If devices are recording footage, this provides a further goldmine of opportunity to deploy deep learning. 365mesh has been able to provide invaluable insights to the transportation industry through integrating AI and deep learning in its IoT solutions:
Identifying types of vehicles, such as cars, taxis and trucks
People and vehicle counting for traffic flow monitoring
Number plate recognition can provide isights into vehicles of interest
Interpreting when a vehicle picks up or drops off a passenger
Identifying driver behaviour, such as illegal parking or unsafe passenger drop offs
According to a Mckinsey study, 75% of experts considered cyber security to be a top priority, but only 16% believed that their company was well prepared for cyber threats.
IoT security cannot be ignored even at the PoC stage. No matter how insignificant an IoT endpoint may seem, it is a point of access into the organisation. There is a famous case where this exact thing happened, criminals took advantage of a seemingly insignificant fish tank sensor and hacked their way into a casino’s network and database.
The more devices and sensors that are connected to the internet, the more potential access points hackers have into a network. A PoC that is not secure, exposes organisations to the risk of financial loss or other damage, and should not be taken lightly.
Hardening, which involves analysing potential threats facing various components, and the development of comprehensive frameworks and guidelines, is achievable if IoT adoption is planned effectively during the early stages. This requires complete control of components within the IoT solution. If the internal task force or external provider can depend on this, then they are well positioned for their security journey.
365mesh carefully reviews all devices, then plan and run test protection practices on each device/sensor to ensure they have the most effective security provisioning. Parent company of 365mesh, Outcomex provides a dedicated team of cyber security practitioners who focus on excellence in governing, deploying, managing, and monitoring security portfolios, ensuring environments are alert to the latest threat profiles, including sensors and IoT solutions.
There is a wealth of opportunity ahead in the IoT space. There are IoT use cases that suit every industry, and it is important that organisations are well-prepared for their adoption journey.
During the early stages or PoC phase in adoption, it’s important to consider long-term goals and a broader scope. Architecture needs to ensure scalability and support the deployment of artificial intelligence and deep learning that will enrich information with valuable insights.
As with any technology program, security is crucial for successful, long-term IoT projects, and it begins with essential control over IoT components.
The 365mesh end-to-end IoT platform provides the required solution components for a mature and secure IoT capability. 365mesh offers industry solutions with use cases and applied AI tailored for unique business needs. Learn more or reach out to the 365mesh team to continue the conversation.