-Most of the emerging software-intensive systems nowadays are very large-scale ones, and inherently sociotechnical. In this position paper, we argue that the peculiar features of such emerging systems (up to millions of interacting components, lacking central control, mixing humans and artificial components) call for novel approaches to coordinate the overall activities and functionalities. Accordingly, we discuss the key challenges to be faced by research in coordination models and technologies, and try to sketch some promising research directions.
The massive diffusion of networked ICT devices
increasingly entangled with our physical and social world, along with
progresses in the area of robotics and AI, are making our
everyday life and economic activities increasingly dependent on
the functionalities of large-scale distributed software-intensive
(i.e., heavily relying on software) systems [
All the above examples, and most emerging ICT-centred
scenarios, are characterised by the innovative services and
functionalities they provide, which require coordination of the
activities of distributed components. However, such systems
exhibit peculiar traits that heavily challenge most of the
coordination models and technologies currently available [
They require the capability of effectively coordinating
the activities of a huge number (up to the millions) of
decentralised autonomous components. This implies
the impossibility of enacting some centralised scheme
of coordination of the activities, as well as the
impossibility of enforcing full control over the activities and
interactions of some (if not most) of the components.
Such systems are inherently heterogeneous and
sociotechnical [
ical and social environments, where their components are required to be context-aware and socially-aware in their interaction, and where consequently any coordination scheme has to be adaptive to the context. In the reminder of our position paper, we detail some of the above-mentioned emerging large-scale socio-technical scenarios, and discuss their peculiar coordination requirements, and how they call for novel models, languages, and infrastructure (Section II). Then, based on the potential impact of such systems on the area of coordination, we identify some of the new challenges and promising directions for theoretical research and technological development (Section III).
The scenarios we outline in this section are generally
representative of some easy-to-envision future scenarios, in
which millions of ICT sensors, actuators, and services will
be called to operate in an orchestrated way, along with the
active contribution of the sensing, actuating, and reasoning
capabilities of humans [
The role of coordination for handling urban traffic has been
widely acknowledged [
First, modern cars already include a number of features to
alleviate driver activities (e.g., the capability of autonomously
park), and fully autonomous self-driving car is going to hit the
streets in a few years [
Second, continuous connectivity to the Internet and the
rise of social networks are enabling a variety of innovative
social mobility models, such as unplanned on-the-fly
ridesharing, dynamic traffic slot allocation, dynamic on-demand
schedule of public transport vehicles [
Critical missions such as armed conflicts and handling
of natural disasters, which typically involve large teams of
soldiers and volunteers, will increasingly involve autonomous
weapon systems and robots as well [
Accordingly, there will be the need of coordinating the activities of possibly large-scale mixed teams of humans and robots in such critical missions. A key issue for coordination in these scenarios comes from the fact that the activities of the system components are situated in environments and situations that can hardly be known in advance, and therefore require the dynamical adaptation of both individual activities and the coordination schemes required to achieve their effective orchestration.
As a consequence, the scenario requires the adoption of
coordination models (and of the corresponding coordination
technologies) that could promote the dynamic and adaptive
expression of different coordination patterns [
Municipalities spend a lot of their money in trying to monitor and maintain our urban environments at the best. Activities include: garbage collection, maintenance of roads and public green, installation and maintenance of public lights, etc. Such activities involve a lot of human work – only a small portion of which calls for specialised skills or tools –, are very costly (due to the need of employ a lot of people), are and typically based on static planning.
Clearly, ICT technologies (such as cameras) make it
possibile in principle to automate some of the sensing activities, and
make them cheaper. However, a more radical evolution that can
be promoted by smart phones and continuous connectivity is
that of dynamically involving citizens in the above tasks, and
make urban management a participatory activity. For instance,
one could dynamically involve citizens to help mapping the
noise level in a town, by having them supply the lack of
appropriate sensors in some parts of the town [
Such kind of participatory urban management activities clearly require bringing together the complimentary capabilities of humans and ICT devices, may involve a very large number of devices and humans, and require involving and coordinating humans in a context-aware way, depending on their current positions, goals, and activities.
Pervasive health care is the new frontier according to a
twofold perspective: the adoption of pervasive models and
technologies to health care systems, and the ubiquitous
availability (everywhere, to everybody) of healthcare [
As a result, coordination in pervasive health care systems generally mandates for robustness, safety, and security: but, first of all, dependability and efficiency even on the large scale are essential to ensure that critical, possibly unplanned operations can be successfully brought to an end successfully. Also, the ability to deal with a huge and inordinately growing amount of knowledge, and to match it in real time with protocols and procedures is a fundamental requirement of pervasive health care systems, which coordination models and technologies are required to address.
III.
In this section we analyse some of the peculiar challenges that arise in large-scale socio-technical systems, such as the one exemplified above, and accordingly point out the main research issues that coordination models and technologies are demanded to address in the next years.
The activities of forthcoming socio-technical systems will involve a variety of agents: humans with mobile devices, ICT sensors and actuators, cameras, self-driving cars, diverse sort of autonomous robots. The features and capabilities of such heterogeneous classes of entities are very different from each other: for instance, one may think at how differently (in terms of modality, timescales, accuracy) humans and artificial vision systems see and classify images, or at how differently robots and humans can assist people.
Accordingly, a coordination language should be expressive enough to enable the representation of coordination schemes among such different classes of components (and, of their associated services and capabilities), yet maintaining a uniform and clear model behind. Similarly, a coordination infrastructure should be able to implement and support such a model in an effective and scalable way, to support the orchestration of myriads of heterogeneous agents physically spread over an urban area. Such issues are so far largely unexplored, and worth being investigated.
One key feature of components and systems in nowadays application scenarios is autonomy, as clearly exemplified by the current transition from human-driven to self-driving cars. Many different acceptations of the term “autonomy”, however, are adopted in different fields – computer science, philosophy, military, psychology, biology, among the many – which lead to different requirements and behaviours for systems and components: and, consequently, for their coordination.
In general, autonomy is typically bound to the existence of
an inner goal, to be self-achieved—by the system as a whole,
or by the component. Since effective coordination usually
requires the mutual understanding of each other goals, the
ability to associate actions in a shared environment with their
motivating goals seems a relevant feature for coordination
models, nowadays. This holds in particular for socio-technical
systems, where humans may easily exhibit mixed, hidden, and
even unaware goals, whereas artificial components could be
either goal-driven or goal-oriented, according to their goal
being implicitly or explicitly represented, respectively [
On the one hand, models of coordination supporting goal /
plan exchange exist (e.g., [
More generally, coordination models will be required to
deal with large numbers of components of many sorts,
featuring many diverse sorts of autonomous behaviours [
When it comes to coordinating the activities of large-scale
decentralised systems of heterogeneous components situated
in dynamic and unpredictable environments, nature may have
something to teach us [
In the past few years, we extensively investigated the
possibility of exploiting nature-inspired approaches as a means
to enable self-organising coordination in context-aware and
situated pervasive computing systems [
However, despite the encouraging results achieved by the SAPERE project, what is still missing is a real assessment of whether nature-inspired coordination model can support the inherent heterogeneity of the emerging systems, and whether they can really scale to support millions of components and large-scale decentralised scenarios. Attacking such an issue would represent a very promising research direction.
A key assumption of most coordination models and languages is that the coordinated components are under the control of engineers, and willing to be coordinated. However, when components belong to multiple stakeholders, or, when such components are humans expected to deliver some service, the effectiveness of the coordination scheme is also related to the effectiveness in incentivising components to participate in the coordination scheme and deliver the necessary services on need.
As far as human are involved, recent work on
persuasive technologies analyses how to induce specific behavioural
changes and persuade people to establish a desired behaviour
[
In any case, there could always be specific classes of
services and behaviours for which persuasive technologies can
hardly apply—e.g., convincing people to park farther than
they would autonomously do, in order to provide for a better
overall parking availability; or, convincing the owner of a
robot to lend it for some time. Therefore, a coordination
model should account for more explicit means to incentivise
participation, such as monetary rewards or social rewards [
The emergence of cloud computing has clearly pointed out how the economic dimension of contemporary ICT services cannot be ruled out even from models. On large-scale sociotechnical systems, money is no longer a secondary concern, which could be deferred after the system design and development: every minimal shift in the costs could produce huge unbalancing in the overall sustainability of systems. And, the availability of certain actions in in given situations is often bound to some measure of cost—differently expressed when actions are by humans (which are typically directly liable for costs) or by ICT components (which are more easily not directly liable)
As far as coordination is concerned, cost issues have generically often been included among the concerns: however, the economic dimension has rarely been explicitly considered. Accordingly, a promising direction for research on coordination is to design models and technologies that explicitly include the economical concerns, possibly associating both individual actions (by either humans or artificial agents) and coordinated activities to their costs (either actual or putative), and making it possible to express coordination policies in terms of economic issues.
The vast availability of data, information, and knowledge,
along with their strong dynamics (e.g., data streams) is a
typical feature of nowadays complex systems—in particular,
large-scale socio-technical systems, where millions of humans
and software agents interact by exploiting and exchanging huge
amounts of data and information. On the one hand, both human
and ICT agents in socio-technical systems typically work
as powerful knowledge sources, accumulating and sharing
information and data, and act according to knowledge and
beliefs, both owned by individuals and shared. On the other
hand, shared environments where coordination activities take
place are typically knowledge-intensive environments, so that
both individual actions and social activities also depend on the
amount, sort, and accessibility of data and information in the
shared environment [
This is why innovative coordination models and technologies will be more and more required to deal with large amount of information and data, to provide support for knowledge modelling and representation, to promote knowledge-rich interactions, and to express knowledge-dependent coordination policies.
CONCLUSIONS
In this position paper we argue that the peculiar features of emerging socio-technical software-intensive systems – up to millions of decentralised situated components, lacking central control, mixing humans and ICT components – call for innovative coordination models and technologies. Accordingly, we discuss the key challenges to be faced by researchers in the area of coordination, and accordingly sketch promising research directions.
As a part of our current research, we are working on
natureinspired coordination models and languages [