How do you make smart machines smarter? That’s the challenge the Industrial Internet, already a fast-growing world of smart machines, currently faces.In a couple of recent editorials,Nikhil Chauhan, Director of Marketing for GE Software, argues that machines need to be more than just smart, they need the capacity to evolve, adapt, and change with the latest developments in technology. The problem today, writes Chauhan, is that each implementation of smart machines and big data solutions is tailored to the narrow needs of particular industries and particular hardware. Any solutions developed aren’t adaptable or scalable to similar problems in other fields. What the Industrial Internet needs, Chauhan says, is a software common tongue.

From Silicon Valley to a Technological Tower of Babel

How does the Industrial Internet go from an assortment of different platforms to a universal ecosystem of applications? According to Chris Murphy, editor of Information Week, the answer lies in leveraging the innovative capital of Silicon Valley startups.In a column on bringing Silicon Valley to big industry, Murphy notes that startups are constantly looking for solutions to problems, whether that means figuring out how to quickly call a cab from your smartphone (like Uber) or devising a way to stream an entire library of movies to wherever you are on whatever device you’re using (like Netflix).

So far, startups have shied away from embracing the Industrial Internet because of the highly specialized knowledge required to understand, let alone solve, the problems of big industry. One potential fix is to have big industries share data more widely. This way, developers can experiment with ideas and invent solutions without actually having to be involved in the day-to-day operations of an industry. What would also help is standardizing the software platforms themselves. According to Chauhan, software needs to decouple from hardware so developers have a common, wide-reaching marketplace to sell their solutions. The good news is that this kind of platform convergence is already starting to happen.

Big software developers are beginning to create large, scalable platforms for the Industrial Internet, particularly in the field of big data analytics. Companies like Pivotal and Belgacom are developing data analysis and machine-to-machine solutions that emphasize universality and scalability. Big data tools like Apache’s Hadoop are robust and adaptable to the demands of many different industries. Efforts to standardize hardware architecture, like Intel’s recently-announced Quark platform, will also help developers to create software that can run on everything from the smallest to the largest scales.

The Power of the Cloud

Common software platforms might be the next phase of expanding the Industrial Internet, but Chauhan is already looking beyond. For the Industrial Internet to truly become a transformative presence, it needs to find a way to keep machines relevant and expandable. How do you keep smart devices like MRI machines from quickly becoming obsolete? The answer is to virtually separate the brain from the body. Chauhan proposes a model for software-defined machines (SDMs), smart machines whosemachine functionality is virtualized in software. These machines willnot onlyhave the ability to execute local applications and be a complete point of presence but the majority of their intelligence will come from cloud-based applications powered bya vast, powerful distributed compute cloud. These SDMs would communicate through high-speed secure networks so that data processing, analysis, and reporting would occur in real time, as though the heavy lifting was being done by the machines themselves. An SDM paradigm, Chauhan declares, is what’s ultimately needed to overcome the complexity and fragmentation present in the Industrial Internet today.

Cloud computing of this kind has already begun in the consumer space. Sony’s Gaikai and Nvidia’s GRID are examples of gaming platforms where almost all the heavy-duty graphics processing occurs in clusters of distributed computers. The rendered output is then simply streamed, in real-time, to the end-user’s console over the Internet.

The main handicap of cloud computing for consumers right now is the relatively low bandwidth of consumer Internet connections. That’s a problem that big industry doesn’t have, with the resources to invest in high speed, secure communication networks.

The Industrial Internet has already seen a lot of success, and it has the potential to grow into a transformative presence of the 21st century. To overcome its outstanding challenges and reach that potential, it needs to effectively adapt successful elements of consumer software platforms and cloud computing to the unique needs of big industry. If it can, we’ll see an Industrial Internet full of automated, self-correcting, self-aware, powerful, and predictive devices – truly smart machines – and all of us will benefit.