Have an account?

  •   Personalized content
  •   Your products and support

Need an account?

Create an account

Seven steps to successful IoT projects

Successful IoT projects start with understanding these seven components of an Internet of Things-based system.

Ask an analyst about the emerging technologies with the greatest impact on the IT market over the next decade, and a common answer will be IoT.

The Internet of Things (IoT) is not a single technology but a collection of technologies, processes, and devices that promises to drive major changes in most organizations. IoT is about using data insights and automation to drive business decisions.

Think about a system that could gather real-time data about a train's pending maintenance while the train is traveling at 200 miles per hour. This could significantly reduce unplanned train maintenance and downtime, trim unexpected costs, provide greater alignment of resources and people, and drive increased revenue and profitability. These kinds of business opportunities indicate why analysts predict the IoT market could reach into the billions or possibly trillions of IoT sensors and devices. Bringing new data efficiency and other data analytics, the IoT market is also predicted to generate trillions of dollars.

The seven aspects of successful IoT projects

In general, IoT systems consist of seven essential stages and technologies. To manage successful IoT projects, you should understand these components.

  1. Devices, sensors, and device management. This part of IoT encompasses the physical sensors and devices that are placed in equipment as well as the centralized tools that maintain those sensors and devices. This includes the ability to update firmware, apply security settings, and so on.
  2. Gateway and edge devices. Sensors and devices must establish some form of communication. To support real-time business technologies and to manage data processing, many initial decisions are made using edge computing tools (a network architecture that enables data processing much closer to the devices and users engaging in these activities) and gateways. The data is then aggregated and sent into an IoT system.
  3. Cold-path analytics. Most organizations lack the insights and knowledge to fully exploit IoT-generated data. To obtain this insight, machine learning (ML) tools analyze a collection of data that has been stored for longer periods of time or data of sufficient volume to identify potential patterns: This is known as cold-path data. Hot-path data involves known patterns, where the appropriate ML and real-time ingestion processes can be applied). When data patterns have been identified, the machine learning algorithms can be published into edge devices and hot-path analytics to support real-time business decisions.
  4. Hot-path analytics. Hot-path analytics allow organizations to apply real-time business algorithms on the sensor data, enabling immediate responses. Often this analysis is done using a set of technologies called complex event processing.
  5. Data storage and advanced analytics. After data has been collected into an IoT system, the data is kept for longer periods of time. This enables additional analysis and insights to be developed. Data may be sent back into cold-path analytics for additional insights or may be used for reporting. Depending on the systems reporting data, this could represent terabytes, petabytes or even exabytes of data.
  6. Line-of-business system integration. IoT systems rarely exist as their own island of technology. Instead, most are connected into line-of-business systems, enabling data insights to directly feed business applications and management systems.
  7. Reporting and dashboards. IoT data will be supplied to reporting and dashboard tools to provide visual indicators of system health, actions happening, or impacts on business.

Not all IoT solutions will use all seven of these stages, and some will jump between stages based on the solution and the business needs. As such, these stages are a general guideline of what projects will require.

The cold-data path and getting started with IoT projects

One of the most common IoT scenarios is that of predictive maintenance—the ability to prevent an equipment failure through proactive management. The concept of predictive maintenance is that if an organization were to know a piece of equipment would fail because of a faulty part, replacing the part in a timely manner could eliminate costly downtime.

While the promise of predictive maintenance exists, most organizations lack the necessary data insights or volume of data. Predictive maintenance requires a significant amount of data to understand failure patterns compared to normal operations. And to compound the challenge, many organizations don’t collect data in timely-enough intervals to be of use by an IoT system.

Even when an organization can collect an adequate amount of data often enough to be valuable, the business may nonetheless lack the insights underlying the algorithms to support complex event-processing systems.

So the most important starting point in an IoT project is the cold-path analytics stage. In this stage, all the data must be loaded into a machine learning solution with a team of data scientists carefully using the data to identify data patterns. This process is iterative: An algorithm will be developed and then tested against the data. Testing will often identify further changes required to properly detect data anomalies. Complicating this further, these steps must be completed for each IoT sensor and device so the health of that part of the equipment is understood.

Once the data patterns are understood and the algorithms created, the ML models can be connected to the hot-path analytics/complex event processing systems and edge computing systems. Edge and hot-path systems with ML models can then apply real-time data insights and alerts.

When getting started with IoT, an organization should begin with a data project. Consider the business goals and objectives. A common mistake is having business goals that are mismatched with the frequency of data gathering. If a goal requires, for example, real-time insights, but the equipment provides data only every few hours, the goal conflicts with operations. Likewise, having only a small amount of historical data isn’t sufficient to create machine learning models to support business goals.

Driving successful IoT projects

Successful IoT projects start with an end goal. IT managers and business executives alike should be able to answer questions such as these:

  • What will an IoT system accomplish?
  • How will the IoT system affect customer experience?
  • What business processes can IoT technologies improve?

Rarely do most organizations have the right data collection, the right sensors, and the right tools in place to meet the business goals. Organizations should start with the data and sensors available as initial starting points and then add additional sensors and devices and expand the data as subsequent steps. This will create the initial cold-path analytics and ML models. These models can then be expanded and enhanced as additional data and sensors become available.

If you’re trying to establish a successful IoT project, consider these questions in the planning and initial pilot phases:

  • What is the business goal of an IoT system and which metrics will identify success?
  • Does the data exist to support the business goal?
  • Are the data patterns already known and understood or is further analysis required?
  • Is there a small or easy proof of concept that can demonstrate business value?

One of the challenges in getting started with IoT technology is getting past the “things” aspect. IT practitioners should look at the specific business problem to be solved. Business goals should also be broken into smaller pieces to demonstrate the value a technology can provide.

Returning to our example of proactive maintenance for a moving train, the ultimate business goal is to prevent maintenance from disrupting service. Solving a specific problem requires many sensors, a significant amount of data, and deep knowledge of every piece of equipment.

If one or two smaller goals can be tested, though, it could establish the business case for continued investment. In the case of the train, identifying a single accelerometer might be valuable to identify operators that provide a smoother riding experience. Identifying just this part of the total train ride and process would open operator training opportunities and improve customer experience. Once an initial business case is proven, additional sensors and “things” can be added to expand insights.

Ultimately, successful IoT projects focus on the business and how IoT can drive innovation. These technologies, combined with processes, policies and analytics, can reduce operational costs, improve safety, and drive profitability while also driving business offers that help the organization stay ahead of the competition.

For more Cisco news:

For more Cisco products and resources:

Sean Bryson

Sean Bryson is a vice president of the Americas Solutions and Innovation business at Hitachi Consulting.