Enabling IoT: Infrastructure and Functional Autonomy

In Part 1, The Internet of Things, we postulated a functional definition of IoT:

The real-time seamless autonomous interaction, via multi-topology networks, of collections of networked elements which may be people, processes, data or things, in accordance with domain-specific, user-defined sets of rules and decision matrices to intelligently achieve operational efficiencies with regard to cost, quality, service and speed.

The assumption inherent in that definition is that two things are required to enable IoT: Infrastructure and Functional Autonomy.

1. Infrastructure: The network, consisting of nodes, connectivity elements, appliances and policies, as well as computing hardware and software
2. Functional autonomy: The domain-specific, user-defined sets of rules and decision matrices required for autonomous interaction.

Infrastructure:

In order to be operationally effective, IoT infrastructure, further, must have two components:

1. Network or communications infrastructure
2. Computing infrastructure.

Network infrastructure: Structurally, IoT is a super-network or collection of networks, and the requirements of network infrastructure are well understood. These relate to nodal identity, location and addressability, connectivity protocols, media, interfaces, aggregation, routing, redundancy, policies, security and so on.

Where IoT infrastructure makes excursions beyond our conventional ideas of networks is at the nodal level where the vast array of sensors and actuators create their own challenges with respect to addressability (think IPv6), authentication, power consumption (think Energy-harvesting Sensors), self-healing ability and so on. Complexity is added – actually, in realty, subtracted, from a network point of view - through sensor fusion, which integrates sensors and microcontrollers to create smart sensors.

However, at the infrastructure level, an IoT network is just that – a network. True, that node-granularity and variety requires the deployment of a plethora of connectivity protocols like NFC, ZigBee, Z-wave, Wireless HART, Bluetooth-LE, WiFi, GSM etc., but the network infrastructure elements remain essentially the same: nodes, access points, aggregators and protocol translators (gateways), switches, routers, firewalls, network appliances and the like.

Computing infrastructure: Largely referred to as the cloud-computing infrastructure, although there are myriad nuances, of privately owned, business leased and potentially public service collections of computer hardware capable of processing the fast streams of massive amounts of data originated by the sensors and trafficked over the networks for the performance of the actuators.

Admittedly, the above description presents a rather simplistic view of many functions that are performed in the computing infrastructure, like visualization, computing, analytics, storage, workload management, scheduling, dynamic provisioning, billing, reporting and other custom runtime tasks designed to ensure agreed upon Quality of Services (QoS), but further discussion is probably beyond the scope of this particular post.

Functional autonomy

Functional Autonomy, which refers to domain-specific rules and decision matrices, introduces a new dimension into the mix – one that requires a combination of two things to address:

1. A thorough understanding of vertical industry or application domain where IoT is being enabled and the application specific parameters that, when optimized and/or automated, constitute and/or unlock value
2. A mechanism (essentially software, although several parts may be possible to be implemented in hardware) to automate the interaction of networked elements based upon the data generated by them, and sets of rules that must govern those interactions.

A close, although somewhat clunky, parallel may be drawn here with Enterprise Resource Planning (ERP) software. ERP software has automated and optimized Supply Chain Management for companies for more than three decades now. It has revolutionized business, unlocking tremendous value by realizing unprecedented efficiencies in procurement, inventory control, manufacturing resource planning, distribution requirements planning, and billing & collections through user-defined sets of rules specific to industry verticals, with minimal user intervention - all the while running on a fairly coarse-grained traditional infrastructure of connected computers. Moderately big data, but hardly fast data.

Functional autonomy mechanisms are somewhat like ERP software, but more fine-grained, with substantially bigger data and much more frequent and faster transaction processing. Functional autonomy mechanisms are also, by design, more 'independent' in their decision making and execution (based on rules, of course) and require far less user intervention and/or approval at each stage of execution.

When Infrastructure and Functional Autonomy come together to seamlessly achieve intelligent operational efficiencies that realize value, it may be argued, per our definition, that IoT has been enabled.

Coming in Part 3: Enabling IoT: A Framework for Vertical Solutions