Lifting the Fog into the Clouds in Future Cities

Share this post:

Dr. Marc Frincu, Associate Professor, West University of Timisoara

There is currently an increasing trend in using the Internet of Things (IoT) in all areas ranging from personal devices to city scale systems. To paraphrase Jim Tully, former VP distinguished analyst at Gartner, the IoT is a revolution waiting to happen and when it finally succeeds it will fundamentally change the way we experience the world. Our daily activities, social lives, and society will change dramatically as the pervasive yet hidden from plain sight IoT will bring more efficiency and less waste.

The IoT will generate Big Data in all its V dimensions (velocity, volume, variety, veracity, value). These data come with great potential, but also with challenges w.r.t. privacy, security, data ownership, and governance. To get an idea of the scale we are looking at, Gartner predicts that by 2020 there will be 25 billion Internet connected things. Understanding the business justification of IoT and knowing what to do with Big Data are key challenges for IoT adoption by enterprises. Once these are understood, more technical aspects will have to be dealt with. However, research cannot wait for the former as it has always been the goal of research to be visionary.

The ubiquitous IoT will challenge existing distributed platforms and will require a redesign of existing software stacks. The main bottleneck prohibiting fast data transmission lies in communication networks. These will be flooded with real-time data from millions of interconnected devices that will have to be sent to cloud for global optimization. These devices will be part of smart entities such as smart grids, smart transportation systems, smart water management systems, and smart health care systems. Together these will catalyze the envisioned smart city and eventually smart world.
New solutions will be needed to enable fast data processing for real-time decisions. Use case examples include fast load control in smart grids relying on volatile renewable sources such as wind and solar and smart transportation systems which automatically optimize traffic based on individual car information, user behavior and city CO2 emissions.

However, two main technical challenges will need to be solved in order to enable the use of these large scale IoT based distributed smart systems. These challenges are driven mainly by the data deluge which will flood networks and make cloud processing unfeasible.

Fog and edge computing

To avoid networks congestion the future IoT will have to enable personal devices such as smartphones to process nearby IoT data by taking into consideration user mobility and job collocation. To complement this approach, edge computing taking advantage of the smart devices’ own computational and storage capacities will be used. In this case, novel low memory low energy data processing algorithms will have to be engineered to be used on dedicated IoT devices like smart meters.
While edge computing or fog computing terms have been around us for some years now to describe the layer at the edge of the network to preprocess data before sending metadata to the cloud, little progress has been made so far to realize this aspect in IoT solutions for real life. The little scalability edge computing offers means that solutions oriented towards fog computing will likely be preferred. In fog computing, fog gateways are used to aggregate and preprocess data and see across multiple devices in a LAN before sending it to the cloud.

Machine learning

To process the Big Data coming from the IoT for critical applications such as energy demand-supply optimization and traffic management, real-time data mining and scalable online machine learning algorithms will be required. These algorithms will rely on novel paradigms such as cognitive computing enabled by cognitive computing systems such as IBM’s Watson. These emerging computational models will have to be adapted for large scale online cloud computing and near edge processing in environments with memory and energy constraints and under real-time data, catalyzing new processing models for variable data streams in dynamic environments where online complex local and global information is required for fast optimization. Cognitive computing will leverage the heterogeneous data coming from a wide array of IoT devices and enable fast intelligent online decisions.

Research driven by strategic projects and common vision

In the past years an increasing emphasis has been put on topics such as a cleaner environment. In our society this has started to be noticed in key sectors including energy, transportation, and water and waste management which have been deemed strategic in countries such as Romania. Energy optimization schemes based on renewables and optimized consumption are devised and enacted in practice; eco-friendly smart and secure transportation systems are implemented in pilot areas; and advanced waste and water management mechanisms are being devised. All these are supported by financing entities such as the European Bank for Reconstruction and Development, and the European Investment Bank which back the funding of EU 2014-2020 programs in R&D and infrastructure. These strategic projects drive research in state of the art IoT through their objectives and current technical limitations.

Universities such as The West University of Timisoara (UVT) which foster some of the largest and advanced IBM HPC infrastructures in Romania, can spearhead research efforts by attracting funds and subsequently engineers and scientists which can work together with established researchers in areas such as distributed computing, artificial intelligence, and smart systems. Companies like IBM can support academic research, and a proper industrial knowledge transfer through key commercial and industrial initiatives in Romania, thus the future IoT could be shaped in these academic institutions.
Romania has a long tradition in adopting the latest technologies in areas such as telecommunications and has a solid IT environment. However, for the IoT vision to fully take shape, more involvement from administrative and industrial parties is needed. Already, pilot systems such as smart grids are being implemented in some areas such as the Transylvania region, however they are non-uniformly applied and depend on the strategic vision and openness of energy utilities’ managers. In addition, the development of IoT in industry alone is driven by economical aspects and by the clients. Hence products are limited to what clients require and understand and what companies are willing to invest in their solutions. To enable the future IoT and spearhead national and regional R&D in the mentioned strategic areas, a more unified and country wide strategic vision for at least the next decade should be sketched and signed by the majority of key players from administration, industry, and academia.

Through a deeper involvement of top Romanian universities such as UVT in industrial R&D by providing expertise in key areas where companies do not have resources or focus, this recipe should provide a solid case for experimental research to be validated and spearheaded into industry fully matured and capable of leading the national and regional IoT front.

Lifting the fog

Concluding, clouds alone cannot handle the upcoming large scale IoT driven smart world. A redefinition of how data is managed and transmitted to the cloud and how edge devices interact to manage local optimizations is required. In addition, the involvement of cognitive computing will enable intelligent decisions at scale and in real time on heterogeneous data from various IoT sensors. These solutions will help lift the fog on the technical difficulties using today’s technologies and bridge edge devices with clouds. However, for all these to take place a common vision and collaboration between administration, industry, and universities is needed. Universities like UVT, hosting large IT infrastructures and expertise in vital areas, will complement the industry in shaping the IoT and can help spearhead the IoT enactment in vital areas at local, regional, and national scale. Together, these aspects will help catalyze Romania’s commitment to strategic projects in areas such as energy, transportation, and water and waste management. The broader impact however will pervade other emerging areas that depend on IoT, examples ranging from smart personalized health care and disaster monitoring and prediction.