Main Authors: Carlos Kamienski, Juha-Pekka Soininen

Adittional Authors: Ronaldo Prati, João Kleinschmidt

Focus Area: 

Internet of Things

Who stands to benefit and how: 

The target audience includes a variety of stakeholders, such as IoT developers and integrators, as well as researchers in different areas and business entrepreneurs.

Position Paper: 

In the current stage of IoT deployment, developers and integrators have to build and deploy the entire end-to-end components of a smart application. In the future, sensing and actuation systems will be already in place, and new developments will have to use these legacy systems, building a complex ecosystem. In this paper, we advocate that openness is a key factor for providing interoperability and facilitating the interaction of new and existing pieces of an end-to-end IoT smart application. This view is instantiated for a smart irrigation scenario with findings and insights coming from the SWAMP project that is building a smart water management platform.

1. Introduction

The Internet of Things (IoT) [1], powered by forthcoming communication technologies (e.g. 5G [2]), and new developments in artificial intelligence (e.g. deep learning [3]), is about tobring a high impact in the way we interact with things, services and people in the next few years. Even though the term IoT has been around for almost 20 years now, currently we are still in the very early stages, with a handful of commercial applications and numerous initial proof-of-concept projects in pilot sites. Full-fledged IoT systems are still not present in our daily lives, with the exception perhaps of some very specific applications, such as traffic management services (e.g. Waze) and ride hailing services (e.g. Uber). In thecurrentstage of IoT deployment, developers and integrators have to build and deploy the entire end-to-end software, hardware and communication infrastructure for providing a smart application. Sensors and actuators must be installed as well as platforms and applications must be developed. In the future, when billions of devices will be connected, sensing and actuation systems will be already in place. In such scenario, new applications will be required to use existing legacy sensing and actuation systems, for streamlining application deployment, backward compatibility and cost savings.This situation will build a complex ecosystem of hardware, software and communication components that will require new IoT smart applications to interact with a variety of other existing and new pieces needed for filingthe gaps and enabling end-to-end smart application deployments to come true. In Europe, the need for ecosystems comprised of platforms and businesses have been already identified by different projects such as IoT-EPI1, BIG IoT2and SmartAgriHubs3. Today, there is a myriad of different protocols, platforms, APIs, and it is not easy to make everything work together, unless everything is developed and deployed from scratch by a single organization [4].In this paper, we advocate that openness is a key factor for providing interoperability and facilitating the interaction of new and existing pieces of an end-to-end IoT smart application. The promotion of collaboration and building a healthy IoT ecosystem will be highly influenced by various levels of openness of the solutions, such as open source, open platform, open services, open data and open knowledge. Also, we claim that the simple and well-known input-process-output(IPO) model of system developmentcan help setting up a new level of understanding of the very concept of IoT, concerned with sensors and actuators. There is no single agreed upon concept of IoT, but all tentative definitions recognize the existence of sensors, whose data isprocessed and generate commands to be executed by actuators aimed at changing the state or behavior of a target system. Currently, there is an understanding of some members of the IoT community that necessarily an IoT solution must work end-to-end, covering the three phases of IoT IPO model: input, process, and output. However, this view is an incomplete picture of the IoT scenario because current projects must usually develop and deploy everything to make it work. We also understand that the process phaseof the IoT IPO is concerned with platforms full of services that interact with each other. In the future ecosystem, clearly understanding the scenario and providing ways to interconnect the pieces is paramount to the success of new smart applications.We exemplify our view with an example of an open IoT ecosystem for smart irrigation coming from the findings and insights gained by the SWAMP project4that is developing and deploying a smart water management platform. SWAMP develops IoT based methods and approaches for smart water management in precision irrigation domain and to pilot the approaches in four places, two pilots in Europe (Italy and Spain) and two pilots in Brazil. Smart Irrigation applicationswill have to interact with different existing sensing and actuation systems, as well as install and operate their own ones. As the end-to-end data flow of distinctive configurations vary, platforms must be prepared to cope with these differences. In the remainder of this paper, section 2introduces the IoT Input-Process-Output model for smart applications, section 3characterizes openness as the key enabler for such applications, section 4briefly introduces the SWAMP Project, section 5introduces our view of an open IoT ecosystem for smart irrigation, and finally section 6draws some conclusions.

2. An IoT Input-Process-Output Model for Smart Applications

In an IoT smart application, data collected by sensors in the field are transmitted to some place where they are processed (usually a cloud, public or private) and, as a result, commands are sent to actuators to change the behavior of the system. Generally speaking, we can identify three major parts or systems that make up the entire end-to-end data flow and its transformation, borrowingfrom the well-known IPO (input-process-output) model:

  • IoT Input System: also known as sensing system, involves data gathering from different sources, mainly from of sensors, but also from external services, such as weather forecast. It encompasses sensingdevices and communication technologies. In other words, IoT Input provides data in a standard protocol (e.g. MQTT) and an agreed data format.
  • IoT Process System: involves a myriad of techniques, methods and algorithms for effective and efficient data storage and processing, aiming at obtaining an improved understanding of the system where the data comes from, and making decisions that will change system behavior automatically.
  • IoT Output System: also known as actuation system, involves the control and change of operational status of the actuation system via commands for actuators. It also involves the use of communication technologiesand devices that perform the desired features.

Smart applications of different types may be modeled as IoT IPO systems. For example:

  • Smart Irrigation: Inputs are soil sensor and weather station data, also weather forecast information; Process is running physical and data-driven (e.g. machine learning) models of water need estimation and irrigation planning; Outputs are commands send to pumps, valves, filters and sprinklers for irrigation.
  • Smart Traffic Management: Inputs are GPS coordinates of vehicles in real time and information fed by users; Process is graph modeling and shortest path algorithms; Outputs are instructions given by the application to the driver (ex. Waze).•Smart Healthcare: Inputs are vital signals and movement patterns, ECG and fall detection; Process is identification of abnormalities compared to normal behavior; Outputs are alarms sent to health professionals to take actions to preserve the physical integrity of patients, such as sending paramedics in an ambulance.
  • Smart Manufacturing: Inputs are operational data coming from machines; Process is comparing them to the normal operation conditions and identification of deviation for the predetermined parameters; Outputs are commands to actuators to reestablish the normal behavior or take any other corrective actions, including stopping machines and sounding alarms.

3. Openness: Key enabler for IoT-based Smart Applications

The promotion of collaboration and building a healthy ecosystem for IoT will be highly influenced by various levels of openness of the solutions. A true open business and research ecosystem must be pursued, based on multiple levels of openness, suchas a) open source software; b) open data; c) open access; d) open innovation; e) open prototypes; f) open experiences; g) open experiments [5]. Whereas some of these levels of openness are widely known, others are still in development. The key challenge here is how to promote such open ecosystem where openness is a pervasive value and stimulates collaboration. Everything should be considered open by design, except for some specific cases such as privacy or intellectual property management [5].Sophisticated IoT Systems dealingwith huge amounts of data coming from sensors generated by IoT-based applications are perfect candidates for business exploitation in different business cases, leveraging from the openness perspective:

  • Open Source (code sharing): The platform maybe be freely shared as open source and business can be developed as consultancy services.
  • Open Platform (platform sharing): In this case, the platform is also offered as a service (Platform as a Service), for organizations that do not want to deal with system operation.
  • Open Services (service sharing): services or microservices shared via well-known and open APIs.
  • Open Data (data sharing): data can be provided as a service under a certain fee for consumers interested in them. In other words, raw data becomes an important asset in the business case, where users connect to the platform and consume data.
  • Open Knowledge (knowledge sharing): Build upon the open data business case, in this open knowledge business case the provider can share not only raw data, but knowledge generated by smart algorithms based on artificialintelligence techniques (e.g. machine learning). Different applications may share different knowledge. For example, a smart irrigation system may share models and results in the form of irrigation prescription maps.

4. The SWAMP Project

TheSWAMP project develops and assesses an IoT-based smart water management platform for precision irrigation in agriculture with a hands-on approach based on four pilots in Brazil, Italy and Spain [6]. The SWAMP Platform can be configured and deployed in different ways thus making up different SWAMP Systems. These systems are customized to deal with the requirements and constraints of different settings, countries, climate, soils, and crops, which requires a good deal of flexibility to adapt to a range of deployment configurations involving a varied mix of technologies.The four SWAMP pilots are based on the similar technical solutions and deal with different crops and have different primary goals:

  • MATOPIBA Pilot (Luis Eduardo Magalhães/Brazil): The Rio das Pedras Farm is located in the MATOPIBA region, and irrigation is mostly performed by center pivots. This main pilot goal is to implement and evaluate a smart irrigationsystem based on Variable Rate Irrigation (VRI) for center pivots in soybean production and save energy used in irrigation.
  • Guaspari Pilot (Espírito Santo do Pinhal/Brazil): The Guaspari Winery transfers the wine grape harvesting to the winter season (June-August) using irrigation techniques. The main goal for Guaspari is improving grapes and wine quality.
  • Intercrop Pilot (Cartagena/Spain): Intercrop Iberica addresses several challenges since production is in a dry area, and a considerable amount of water comes from a desalination plant. The primary goal for Intercrop is using water more rationally, since water supply is not guaranteed.
  • CBEC Pilot (Bologna/Italy): The Consorzio di Bonifica Emilia Centrale (CBEC) pilot provides water for about 5,000 farms, via a large network of canals, using water taken from rivers. Currently, a great deal of water provided to farmers is not effectively used for irrigation due to a variety of losses along the way. The main goal of the CBEC pilot is optimizing water distribution to the farms.

5. An Open IoT Ecosystem for Smart Irrigation

The SWAMP project is the source of inspiration for our view of an Open IoT Ecosystem, which can be the basis of any smart application, although in our case it is particularly applied to irrigation in agriculture. The experience with the development of the SWAMP Platform for the four pilots has been bringing us important insights of challenges and how they should be addressed to foster the development of IoT and provide a faster time to market.Currently, a good deal of researchers, developers, integrators and practitioners in general consider that necessarily an IoT solution must work end-to-end, covering the three phases of IoT IPO model: input, process, and output. However, this view is an incomplete picture of the IoT scenario, because at these early days of IoT, deployments usually find no infrastructure in place and therefore everything must be developed from scratch. In fact, there might be a wealth of different ways of implementing IoT systems and not all of them involve the direct access to both sensors and actuators.As the deployment of IoT Smart Applications become more common, in the near future legacy sensing (input), platform (process), and actuation (output) systems will be already installed. Obviously, this might be seen as both a benefit and a drawback. A benefit because the existing infrastructure accelerates the deployment of new smart applications. A drawback because legacy systems frequently impose limitations to the offering of modern technologies.Therefore, in a truly Open IoT Ecosystem smart applications will need to interact with different existing and new system in the three phases of the IPO model. Sometimes the existing sensing and actuation systems will belong to the platform developer/integrator that will be able to improve it. Also, sometimes it will belong to third-party providers and the interfacing will need to be via well-known and agreed-upon APIs, most likely running in public or private clouds. In other words, platform developers (such as the SWAMP project members) will have to address both concerns of directly dealing with sensors and actuators, as well as to existing sensing and actuation systems, treated as external systems.Figure 1depicts different possibilities for an end-to-end IoT Smart Application to deal with the three phases of the IPO Model. In the center we represent the SWAMP approach for the Intercrop pilot, also currently the most common one, where the platform must be developed as well as sensors and actuators must be controlled directly by the platform. However, other combinations are possible. For the MATOPIBA pilot, SWAMP manages the sensors to collect data (top left picture), processes it in the platform and interacts with an existing irrigation system for output provided by Focking5. For the Guaspari pilot, SWAMP also manages sensors and the platform, and interacts with the irrigation system provided by Netafim6. The CBEC pilot is not represented in Figure 1as it is focused on water distribution. Figure 1also illustrates different interactions in an Open IoT Ecosystem. The bottom strip depicts an end-to-end system where sensing, platform and irrigation belong to the same company. Furthermore, there might a variety of platforms that provide and use different services among each other. For example, irrigation services may be provided by SWAMP to other IoT Platforms.Figure 2and Figure 3depict end-to-end smart irrigation scenarios to further illustrate the working of two different types of interactions of the Open IoT Ecosystem. They present simplified views of an IoT infrastructure deployment for smart irrigation, consisting offour locations following an IoT computing continuum, composed of Device (sensors and actuators), Mist (field nodes such as gateways), Fog (farm on-premise computing infrastructure) and Cloud (data storage and processing place). The four instances of this continuum define the end-to-end data path starting with data collected by sensors up tocommands executed by actuators.

Figure 2 illustrates a scenario where input (sensing), process (platform) and output (irrigation) belong to the same organization, which is currently the most common deployment strategy. The numbers in blue circles represent a simplified sequence of the end-to-end data path. Soil moisture sensors send data via LoRaWAN to the gateway installed in the Mist node (1). A weather station also sends data to the Mist node, but via a serial wired interface (1).From there, the Mist node forwards data packets to the Fog installed in the farm office where mostly IoT communications and data checking functions are performed (2). From there, data is sent through the Internet (4G) directly to the IoT Platform (e.g. FIWARE) in the cloud (3), passing it to irrigation specific components (4), represented as Irrigation Planning that abstracts the functions of water need estimation based on crop and soil information and irrigation optimization that generates an optimized and real prescription that is aware of different physical and financial constraints (5). Farmers are shown the irrigation prescription via the Farmer App (5) and approve or change it, which in turn is sent back to Irrigation Operation (6) that controls the irrigation system. From there, irrigation commands follow the way back to the Mist going through the IoT Platform (7) and Internet (8) and Fog (9). Finally, irrigation commands reach sprinklers, pumps and valves (10).

Figure 3 is similar to Figure 2in the platform phase, but differs in the way the interaction with sensors and actuators is performed. In this case, there is a sensing and an irrigation  system installed and the platform interacts with them via APIs that provide services running in different clouds. Notice that the Open IoT Ecosystem allows a variety of different interaction patterns, and Figure 3represents only one possibility in the specific case where input (sensing), process (platform) and output (irrigation) belong to different organizations. Also, Figure 3does not contain a fog node, since the decision regarding its deployment is up to the of sensing and irrigation infrastructure providers.

6. Conclusion

An open IoT ecosystem will allow developers and integrators to navigate in the myriad of technologies, platforms, solutions andexisting infrastructure when developing smart applications in different areas such as farming, cities, healthcare and industry. Dividing up an end-to-end view of an IoT-based smart application into input, process and output phases may be useful to provideadditional understanding of the big picture. Our experience with the development a smart irrigation solution within the SWAMP Project has taught us so far that these two factors are key for building a thriving open ecosystem for smart irrigation.

7. References

  • [1]Atzori, L., Iera, A., Morabito, G., "The Internet of Things: A survey", Computer Networks, 54(15), October 2010.
  • [2]Agiwal, M., Roy, A., Saxena, N., “Next generation 5G wireless networks: A comprehensive survey”, IEEE Communications Surveys & Tutorials, 18(3), 2016.
  • [3]LeCun, Y., Bengio, Y., Hinton, G., “Deep learning”, Nature, 521(7553), 2015.
  • [4]Kamienski, C., Jentsch, M.; Eisenhauer, M., Kiljander, J., Ferrera, E., Rosengren, P., Thestrup, J., Souto, E., Andrade, W., Sadok, D., “Application Development for the Internet of Things: A Context-Aware Mixed Criticality Systems Development Platform”, Computer Communications, 104, May 2017.
  • [5]Kamienski, C., Soininen, J.P., Fernandes, S. "What will the future hold for EU-BR collaboration in ICT?", Workshop on Cloud Networks (WCN 2018), Natal / Brazil, July 2018.
  • [6]Kamienski, C., Soininen, J.-P., Taumberger, M., Dantas, R., Toscano, A., Salmon Cinotti, T., Filev Maia, R., Torre Neto, A., “Smart Water Management Platform: IoT-Based Precision Irrigation for Agriculture”, Sensors, 19(2), January 2019.