Advanced technologies such as distributed ledgers and AI methods of analysis and machine learning allow for increasingly intelligent and distributed automation and control of industrial processes, altering traditional notions of closed, centralized, statically defined and functionally isolated automation systems. The recent paper Control in DIO discusses the role blockchains can play in the automation of value production in decentralized industrial systems.
Blockchain technology has a significant potential role to play in the automation of industrial processes. In addition to its role in implementing cryptocurrencies (e.g., Bitcoin) and smart digital contracts, blockchains support the operation of decentralized industrial organizations (DIO), establishing economic value derived from advanced industrial automation (IA) applications—often in the context of emerging Cyber-Physical (CPS) and Industrial Internet of Things (IIoT) systems.
This paper provides a brief introduction.
The proliferation of interconnected IoT devices and cyber-physical systems (CPS) poses significant and potentially serious concerns for safety, security, privacy and integrity.
For example, take the matter of user credentialing and authentication. Consider systems that employ passwords and two-factor authentication to identify users and interconnected systems. Typically, identifiers in the form of cell phone numbers or email addresses are used to gain access to system services. These are volatile and subject to loss, cancellation or theft, resulting in a temporary or permanent loss of service – a system fault.
The occurrence of an authentication fault in one system may cause a cascade of faults in the web of interconnected and independently produced devices and systems, posing a challenge to identifying the source of the fault and its remediation. For safety-related applications, or applications that have cost penalties for loss of service, such authentication failures can be disastrous.
Authentication failures is a class of system fault that can occur in distributed systems. The effects of such failures can be minimized in systems with integrated designs. However, they are nearly impossible to predict a prior in heterogenous systems separately designed and deployed over time and in ad hoc ways. Importantly, accountability for failure detection, diagnosis and response in heterogenous systems may be difficult to establish.
The domain of cybersecurity includes many classes of potential faults, including authentication, data integrity, safety, system integrity (e.g., performance) and availability. These are system specification and design concerns, alone and in combination. In IoT and CPS environments they are especially challenging.
Governance Systems provide services for regulating the states and behaviors of physical or synthetic value production processes as formal elements of cyber-physical systems .
This whitepaper discusses Cognition in relation to Stratum 4’s technical framework for intelligent automation systems.
Premise: innovation in a digital economy requires treating value production in an enterprise, regardless of its mission and scope, as a computation.
Due to the effects of engaged human beings there is ambiguity, and therefore clarity, in equating “enterprise” with a “quantifiable unit of computation.” It is caused partly semantic, the meaning of enterprise, and partly in the belief that we can automate whatever can be exactly specified, especially the precise meaning of value production.
In any social group there are several possibly competing definitions of the term enterprise. For some it is a noun, signifying an organizational unit (e.g., commercial, religious or political) of some unspecified size and mission. For others, it is an adjective, modifying a noun, such as “Mary is an enterprising person.” And still others might associate the term with a project or process, a verb signifying a specific kind of action.
Implicitly underlying these interpretations is the concept of a performance metric, a value proposition, defining the benefits of an organizational unit, achievements derived from Mary’s industriousness, or results achieved by a process. Value propositions are the cause celebre that justify an enterprise’s existence, the benefits of being enterprising, and the achievements of enterprise activities.
Value propositions are predicates, conditional statements of the form
if <condition>, then <perform action A>, else <perform action B>
Value propositions are computable expressions, defined in possibly nested “if…, then…, else…” statements, and implemented in neural networks, Petri nets, fuzzy logic, mathematical or algorithmic software, firmware, wetware or hardware expressions.
Essentially all value production is a computation, whether in the form of an enzyme catalyzing a biochemical reaction or a multinational company offering a new commercial service for a customer in one of its regional market segments. Consequently, we can think in algorithmic terms and study the dimensions and processes of value creation. With availability of high performance computing and associated modeling, simulation, data analytics and visualization tools, the dawn of the 21st century is witnessing uprecidented growth in our understanding of physical and synthetic sciences (e.g., physics, biology, chemistry, economics, medicine, environment), developing their representative computational sciences as enablers supporting new insights and discoveries.
Admittidly, some computations define value propositions that are determinsitc (predictable), where others, perhaps the most important ones, are nondeterminstic and governed by probablistic or chaotic effects. Nevertheless, attempts to provide automation of these computations, intelligent or otherwise, requires expressing them with as much fidelity as our computational sciences permit.
Jay Bayne, PhD
Principal, Stratum 4
Research Professor, Mathematics, Statistics & Computer Science, Marquette University