Much of the population uses the Internet for information and communication; however, recent developments have introduced another Internet model. Everything from home monitoring to automatic vehicle control is available through sensors, controllers and software involved in a machine-to-machine (M2M) Web. This Internet of Things (IoT) is the next major driver of online applications and cloud services. IoT cloud models can be divided into...
three classes: sensor cloud, control cloud and analysis cloud. These could all become targets for private cloud design and public cloud services.
Although it's interesting to think of a world with billions of sensors being read by applications and users, the privacy concerns and sheer magnitude of finding and interpreting sensor data make it unlikely. Users would be exposed to malicious tracking and intrusions into their activity, and companies would find governance principles impossible to apply. Even the sensors could be plagued with malicious distributed denial-of-service (DDoS) attacks or have too many people polling for information.
Cloud models could buffer critical IoT elements from direct access, as well as make information and control options into easily accessible services for users. The IoT cannot succeed as a disorderly cloud of sensors with public access. However, it can succeed as a collection of cloud services because we're already learning how to deploy, exploit and secure that environment. The cloud will not displace IoT, it will enhance it.
Using sensor cloud models
The sensor cloud may be the most appealing opportunity for a software as a service (SaaS) IoT. Any broadly based cloud provider could offer sensor cloud facilities, but the ISPs and telcos have the most buyer credibility in that space. SaaS for sensor cloud could be a launching point for other IoT cloud services. It would also drive competition and increase the adoption rate for IoT overall.
In a realistic IoT model, distributed cloud apps read sensors and store information in databases for secure user access. Hadoop is an obvious choice for IoT sensor data storage, where data associated with location, sensor type and output interpretation information would be possible to scan and interpret for information. However, the database model of the sensor cloud would not be appropriate for latency reasons because some IoT apps are designed for real-time control. Flow services -- such as Amazon's Flow Framework -- could be a source of raw sensor information and an input to a database where sensor information is stored for non-real-time analysis.
Keeping a lid on the control cloud
IoT controls are network elements that can change the behavior of physical systems. A command sent to a control element can turn a traffic light red or green, open a gate, sound an alarm and more. Clearly, controls have more privacy constraints than sensors. Most would never have public access and others would require a high degree of security for public safety.
The combination of sensor and control clouds resembles a management information base (MIB) commonly used to represent status and parameter control of routers and servers. Cloud applications could write to a variable used to change the state of the control. Infrastructure to Application Exposure (i2aex) is an Internet Engineering Task Force, or IETF, proposition for a repository-based framework for controlling network devices from applications. Network applications that read and manipulate MIB data could be used with the IoT.
Should a control cloud allow access to control points directly? Or should it provide a software gateway point that could provide security? The latter could perform any logical transformation of formats needed for sensor events to flow to control points directly for industrial and other real-time processes. Flow-based mechanisms for sensor event handling that are adopted in the cloud could be extended to allow control software components to be coupled into the flows. However, many believe that control cloud applications should be exposed as analysis cloud components.
Analyzing IoT data with an analysis cloud
The IoT analysis cloud is a set of services that correlates or analyzes data to reach useful conclusions beyond digesting sensor data. For example, the IoT used on traffic and control signals for emergency vehicle movement would be ideal for finding the best route for ambulances or fire trucks, based on sensor data and available signal control points. Analysis can also be used to avoid revealing private information. A service could suggest a meeting point for friends without revealing their current locations.
The IoT analysis cloud is SaaS by nature and can be used as service-oriented architecture (SOA) processes or REST resources. Control cloud elements could be similarly used, and all database management services can be modeled RESTfully as well.
The IoT could be considered an Internet of cloud components, representing a new population of sensor and control devices. In addition to better, faster and more consistent implementations, the IoT cloud model could unite Internet of Things and cloud security mechanisms to improve security in both spaces.
About the author:
Tom Nolle is president of CIMI Corp., a strategic consulting firm specializing in telecommunications and data communications since 1982.
Cloud trumps traditional IT systems for security
Cloud security tools come up short against shadow IT
How has the Internet of Things evolved?
Dig deeper on Data security in the cloud