A simple case study – representative of a larger class of emerging pervasive scenarios – can help grounding our
I. INTRODUCTION arguments and sketching the requirements of future pervasive
The ICT landscape, yet notably changed by the advent of services.
ubiquitous wireless connectivity, is further re-shaping due to It is a matter of fact that we are increasingly surrounded by
the increasing deployment of pervasive computing technolo- digital displays: from those of wearable devices to wide
wallgies. Via RFID tags and alike, objects will carry on digital mounted displays pervading urban and working environments.
information of any sort. Wireless sensor networks and camera Currently, the latter are simply conceived as static information
networks will be spread in our cities and buildings to monitor servers to show information in a manually-configured manner
physical phenomena. Smart phones and alike will increasingly – e.g., cycling some pre-defined commercials or general
intersense and store notable amounts of data related to our personal, est news – independently of the context in which they operate
social and professional activities, other than feeding (and being and of the users nearby. However, such displays infrastructures
fed by) the Web with spatial and social real-time information can be made more effective and advantageous for both users
[
This evolution is contributing to building integrated and and adaptable information service infrastructures.
dense infrastructures for the pervasive provisioning of general- First, information should be displayed based on the current
purpose digital services. If all their components will be able to state of the surrounding physical and social environment. For
opportunistically connect with each other, such infrastructures instance, by exploiting information from surrounding
tempercan be used to enrich existing services with the capability ature sensors and from user profiles, an advertiser could have
of autonomously adapting their behavior to the physical and ice tea commercials – instead of liquor ones – being displayed
social context in which they are invoked, and will also in a warm day and in a location populated by teenagers. Also,
support innovative services for enhanced interactions with the actions could be coordinated among neighboring displays, e.g.,
surrounding physical and social worlds [
We are already facing the commercial release of a variety infrastructure itself, deeply embedded in the physical of early pervasive services trying to exploit the possibili- space, should effectively deal with spatial concepts and ties opened by these new scenarios: GPS navigator systems data. providing real-time traffic information and updating routes Second, and complementary to the above, the display infrasaccordingly, cooperative smart phones that inform us about tructure and the services within should automatically adapt to their own modifications and contingencies in an automatic way without experiencing malfunctionings, and possibly taking advantage of such modifications. Namely, when new devices are deployed or when new information is injected, a spontaneous re-distribution and re-shaping of the overall displayed information should take place. For instance: the deployment of a big advertising display in a room may suggest re-directing there all the ads previously forwarded to the personal displays of users; the injection of a new information service could induce aggregating it with the existing ones to provide a more complete yet uniform service. In terms of a general requirement for decentralized and dynamic scenarios:
should inherently exhibit properties of autonomous adaptation and management, to survive contingencies without human intervention and at limited costs.
Third, the display infrastructure should enable users – other
than display owners – to upload information and services to
enrich the offer or adapt it to their own needs. For instance,
users may continuously upload personal content (e.g., pictures
and annotations related to the local environment) from their
own devices to the infrastructure, both for better
visualization and for increasing the overall local information offer.
Similarly, a group of friends can exploit a public display
to upload software letting it host a shared real-time map to
visualize what’s happening around (which would also require
opportunistic access to the existing environmental sensors and
to any available user-provided sensors, to make the map alive
and rich in real-time information). In general, one should
enable users to act as “prosumers” – i.e., as both consumers
and producers – of devices, data, and services. Not only
this will make environments meet the specific needs of any
specific user (and capture the long tail of the market), but
will also induce a process of value co-creation increasing
the overall intrinsic value of the system and of its services
[
Prosumption and Diversity — The infrastructure should tolerate open models of service production and usage without limiting the number and classes of services provided, and rather taking advantage of the injection of new services by exploiting them to improve and integrate existing services whenever possible, and add further value to them.
Finally, beside short-term adaptation, in the longer-term any pervasive infrastructure will experience dramatic changes related to the technology being adopted, as well as in the kinds of services being deployed and in their patterns of usage. For instance, a display infrastructure will somewhen integrate more sophisticated sensing means than we have today and will possibly integrate enriched actuators via which to attract user attention and interact with them. This can be the case of personal projection systems to make any physical object become a display, or of eyeglass displays for immersive perception and action. While this can open up the way for brand new classes and generation of services to be conceived and deployed, it also requires that such evolution can be gradually accommodated without harming the existing infrastructure and services. As a general requirement:
Eternity — The infrastructure should tolerate long-term evolutions of structure, components, and usage patterns, to accommodate the changing needs of users and technological evolution without forcing significant and expensive re-engineering efforts to incorporate innovations and changes.
NATURE-INSPIRED PERVASIVE SERVICE ECOSYSTEMS
Could the above requirements be met by architecting
pervasive service environments around standard service-oriented
architectures (SOA) [
In general, SOA consider the inter-related activities of service components to be managed by various infrastructural (middleware) services such as: discovery services to help components get to know each other; context services to help components situate their activities; orchestration services to coordinate interactions according to specific application logics; and shared dataspace services to support data-mediated interactions. To architect a pervasive display environment in such terms (Figure 1-up), one has to set up up a middleware server in which to host all the necessary infrastructural services to support the various components of the scenario, i.e., displays, information and advertising services, user-provided services, sensing devices and personal devices.
Such components access the discovery service to get aware of each other. However, in dynamic scenarios (users and devices coming and going), components are forced to continuously access (or being notified by) the discovery service to preserve up-to-date information—a computational and communication wasting activity. Also, since discovery and interactions among components have to rely on spatial information (i.e., a display is interested only in the users and sensors in its proximity), this requires either sophisticated context-services to extract the necessary spatial information about components, or to embed spatial descriptions for each component into its discovery entry, again inducing frequent and costly updates to keep up with mobility.
To adapt to situations and contingencies, components should
be able to recognize relevant changes in their current
environment and plan corrective actions in response to them, which
again require notable communication and computational costs
for all the components involved. Alternatively, or
complementary, one could think at embedding adaptation logics into some
specific server inside the middleware (e.g., in the form of
autonomic control managers [
To reduce the identified complexities and costs and better match the characteristics of the scenario, one could think at a more distributed solution, with a variety of middleware servers deployed in the infrastructure to serve, on a strictly local basis, only a limited portion of the overall infrastructure. For instance (Figure 1-bottom), one could install one middleware server for each of the available public displays. It will manage the local display and all local service components, thus simplifying local discovery and naturally enforcing spatial interactions. Adaptation to situations is made easier, thanks to the possibility of recognizing in a more confined way (and at reduced costs) local contingencies and events, and of acting locally upon them.
With the adoption of a distributed solution enforcing locality, the distinction between the logics and duties of the different infrastructural services fades: discovering local services and devices implies discovering something about the local context; the dynamics of the local scenarios, as reflecting in the local discovery tables, makes it possible to have components indirectly influence each other (being their actions possibly dependent of such tables), as in a sort of shared dataspace model. This also induces specific orchestration patterns for components, based on the local logics upon which the middleware relies to distribute information and events among components and to put components in touch with each other.
A problem of this distributed architecture is to require solutions both to tune it to the spatial characteristics of the scenario and to adaptively handle contingencies. The logics of allocation of middleware servers (i.e., one server per display) derives naturally only in a static scenario, but the arrival and dismissing of displays requires the middleware servers to react by re-shaping the spatial regions and the service components of pertinence of each of them, also correspondingly notifying components. To tackle this problem, the actual distribution of middleware servers should become transparent to service components—they should not worry about where servers are, but will simply act in their local space confident that there are servers to access. Moreover, the network of servers should be able to spontaneously re-organize its shape in autonomy, without directly affecting service components but simply adaptively inducing in them a re-organization of their interaction patterns.
Pushed towards a very dense and mobile network of nodes and pervasive devices, the architecture will end up being perceivable as a dense distributed environment above which a very dynamic set of spatially-situated components discover, interact, and orchestrate with each other. This is done in terms of a much simplified logics, embedded into the unique infrastructural service, subsuming the roles of discovery, context, dataspace, and orchestration services, and taking the form of a limited set of local rules embedded in the spatial substrate itself. That is, we would end up with something that notably resembles the architecture of natural ecosystems: a set of spatially situated entities interacting according to welldefined set of natural laws enforced by the spatial environment in which they situate, and adaptively self-organizing their interaction dynamics according to its the shape and structure.
Going further than architectural similarity, the natural
metaphor can be adopted as the ground upon which to rely to
inherently accommodate the requirements of pervasive service
scenarios. Situatedness and spatiality are there by construction.
Adaptivity can be achieved because of the basic rules of the
game: the dynamics of the ecosystem, as determined by the
enactment of laws and by the shape of the environment, can
spontaneously induce forms of adaptive self-organization
beside the characteristics of the individual components.
Accommodating new and diverse component species, even towards a
long-term evolution, is obtained by making components part
to the game in respect of its rules, and by letting the ecosystem
dynamics evolve and re-shape in response to the appearance
of such new species. This way, we can take advantage of
the new interactional possibility of such new services and
of the additional value they bring in, without requiring the
Fig. 2. A Conceptual Architecture for Pervasive Service Ecosystems
individual components or the infrastructure itself (i.e., its laws
and structure) to be re-engineered [
Indeed, nature-inspired solutions have already been
extensively exploited in distributed computing [
The lowest level is the concrete physical and digital ground on which the ecosystem will be deployed, i.e., a dense infrastructure (ideally a continuum) of networked computing devices and information sources. At the top level, prosumers access the open service framework for using/consuming data or services, as well as for producing and deploying in the framework new services and new data components or for making new devices available. In our case study, they include the users passing by, the display owners, and the advertising companies interested in buying commercial slots. At both levels openness and its dynamics arise: new devices can join/leave the system at any time, and new users can interact with the framework and can deploy new services and data items on it. In our case study, we consider integration at any time of new displays and new sensors, and the presence of a continuous flow of new visualization services (e.g. commercial advertisers) and users, possibly having their own devices integrated in the overall infrastructure.
In between these two levels, lay the abstract computational components of the pervasive ecosystem architecture. Species — This level includes a variety of components, belonging to different “species” yet modeled and computationally rendered in a uniform way, representing the individuals populating the ecosystem: physical and virtual devices of the pervasive infrastructure, digital and network resources of any kind, persistent and temporary knowledge/data, contextual information, software service components, or personal user agents. In our case study, we will have different software species to represent displays and their displaying service, the various kinds of sensors distributed around the environment and the data they express, software agents to act on behalf of users, display owners, and advertisers. In general terms, an ecosystem is expected to be populated with a set of individuals physically deployed in the environment, situated in some portion of the ecosystem space, and dynamically joining/leaving it.
Space — This level provides and gives shape to the spatial fabric supporting individuals, their spatial activities and interactions, as well as their life-cycle. Given the spatial nature of pervasive services (as it is the case of information and advertising services in our case study), this level situates individuals in a specific portion of the space, so that their activities and interactions are directly dependent on their positions and on the shape of the surrounding space.
Practically, the spatial structure of the ecosystem will be reified by some minimal middleware substrate, deployed on top of the physical deployment context, supporting the execution and life cycle of individuals and their spatial interactions. From the viewpoint of such individuals, the middleware will have to provide them (via some API) with the possibility of advertising themselves, accessing information about their local spatial context (there included the other individuals around), and detecting local events. From the viewpoint of the underlying infrastructure, the middleware should provide for transparently absorbing dynamic changes and the arrival/dismissing of the supporting devices, without affecting the perception of the spatial environment by individuals.
Technologically, this can be realized by a network of
active data-oriented and event-oriented localized services
(e.g., tuple spaces [
Eco-Laws — The way in which individuals (whether services components, devices, or generic resources) live and interact is determined by the set of fundamental “ecolaws” regulating the ecosystem model. Enactment of ecolaws on individuals will typically affect and be affected by the local space around and by the other individuals around. In our case study, eco-laws might provide at automatically and dynamically determining to display a specific information on a screen as a sort of automatic reaction to specific environmental conditions, or at having two displays spontaneously aggregate and synchronize with each other in showing specific advertisements. Although the set of eco-laws is expected to be always the same for a specific implementation, they will possibly have different effects on different species. Accordingly, their enactment may require the presence of some meaningful description (within the uniform modeling of individuals) of the information/service/structure/goals of each species and of their current context and state. These descriptions, together with proper “matching” criteria, define how the eco-laws apply to specific species in specific conditions of the space. We emphasize that, in the proposed architecture, the concept of “semantic description” of traditional SOA to facilitate discovery turns into a concept of “alive semantic description” (dynamically changing as the context and state of components change), to properly rule the dynamic enactment of eco-laws. Practically, to be able to code eco-laws and enact them, the middleware substrate should proactively mediate inter-component interactions, and act as an active space in which to store their continuously updating semantic descriptions, so as to adaptively support the matching process triggering eco-laws in dependence of the current conditions of the overall ecosystem. Technologically, since most tuple-based middleware systems are currently enriched with the capability of reacting to events and of configuring the matching process, they could well act as the interaction media in which to embed and enact ecolaws.
The proposed architecture represents a radically new perspective on modeling service systems and their infrastructures. The typically heavyweight and multifaceted layers of SOA are subsumed by an unlayered universe of components, all of which underlying the same model, living and interacting in the same spatial substrate, and obeying the same eco-laws—being the latter the only concept hardwired into the system.
This rethinking is very important to ensure adaptivity, diversity, and long-term evolution: no component, service or device is there to stay, everything can change and evolve, self-adapting over space and time, without undermining the overall structure and assumptions of the ecosystem. That is, by conceiving the middleware in terms of a simple spatial substrate in charge of enforcing only basic interaction rules, we have moved away from the infrastructure itself the need of adapting, and fully translated this as a property of the application level and of its dynamics.
The dynamics of the ecosystem will be determined by individuals acting based on their own goals/attitudes, yet being subject to the eco-laws for their interactions with others. Typical patterns that can be driven by such laws may include forms of adaptive self-organization (e.g., spontaneous service aggregation or service orchestration, where the eco-laws plays an active role in facilitating individuals to spontaneously interact and orchestrate with each other, also in dependence of current conditions), adaptive evolution (changing conditions reflect in changes in the way individuals in a locality are affected by the eco-laws) and of decentralized control (to affect the ecosystem behavior by injecting new components in it).
Beside the above architectural guidelines, what actual shape
can species, space, and eco-laws take in an actual
implementation? Identifying and validating specific solutions in this
direction will be a key challenge of pervasive computing in
the next years. Yet, we argue that whatever solution will most
likely get inspiration from one of the key natural metaphors
already explored in the literature, or possibly extract specific
desirable aspects from many of them towards a new
synthesis. Key metaphors include physical [
All the metaphors, by adhering to the proposed architecture, are by construction spatially situated, adaptive by selforganization, and open to host diverse and evolving species. However, when it comes to modeling and implementing, different metaphors may tolerate with variable efficiency and complexity the enforcement of adaptive self-organization patterns and the support of diversity and evolution. In addition, since the service ecosystem is here to ultimately serve us, it is necessary to analyze how and to which extent the metaphors facilitate exerting forms of decentralized control over the ecosystem behavior, in order to direct its selforganizing activities and behavior and not to lose control over it. Ideally, a metaphor should be able to support these features while limiting the number and complexity of ecolaws, the complexity of individuals and their environment, while keeping the infrastructure lightweight and the overall execution efficient.
These consider species as sort of computational particles,
living in a world of other particles and virtual computational
force fields, the latter acting as the basic interaction means.
Activities of particles (to be practically modeled and
implemented as reactive agents) are driven by laws that determine
how particles spread fields, how fields propagate and reshape
upon changing conditions, and how they influence particles
(those whose semantic description “matches” some criterion).
Particles change their status based on the perceived fields,
and move or exchange data by navigating them (i.e., particles
spread sort of data particles to be routed according to the shape
of fields). The world in which such particles live and fields
spread and diffuse can be either a simple (euclidean) metric
world mapped in the physical space, or a virtual/social space
mapped on the technological network. From the infrastructural
viewpoint, a network of local tuple spaces will have to
proactively support the storing of local field values, the propagation
and continuous update of fields across the network, and the
notifications about these changes to individuals. For instance,
the TOTA middleware [
In the case study, we can imagine display services as masses emitting gravitational-like fields (in the form of broadcast events or spanning trees over the network). Such fields have different “flavors” (i.e., different semantic descriptions, reflecting the characteristics of users around and the environment conditions) and an intensity proportional to either their dimension or the available display slots. Information and advertiser agents can behave as masses attracted by fields with specific flavor, eventually getting in touch with suitable displays for their information and ads. Upon changing conditions, the structure and flavors of diffused fields will change, providing for dynamically re-assigning information and ads to different displays.
Physical metaphors have been extensively studied for their
spatial self-organization features, and for their effectiveness
in facilitating the achievement of coherent behaviors even
in large scale systems—for load balancing, data distribution,
clustering, aggregation, and differentiation of behaviors. The
conceptual tools available for controlling the spatial behavior
and the dynamics of such systems are well-developed, most
of them related to acting on how fields propagate and
dynamically change—by which it is actually possible to exert
control over the overall system behavior. On the other hand,
such metaphors hardly tolerate high diversity and evolution.
In fact, to support very diverse species and behaviors (at a
time and over time), eco-laws must become complex enough
to tolerate a wide range of different fields and propagation
rules, with an increase in the complexity of the model and
in burden on the infrastructure [
These consider species as sorts of computational
atoms/molecules (again modeled and implemented as
reactive agents), enriched with semantic descriptions acting
as the computational counterpart of the bonding properties
of physical atoms/molecules (yet made dynamic to reflect
the current state and context of individuals). Accordingly,
the laws that drive the overall ecosystem behavior resemble
chemical reactions, that dictate how chemical bonding
between components take place (relying on some forms of
pattern matching between semantic descriptions), and lead
to aggregated, composite, and new components, and also to
growth/decay of species. The world where individuals live
is typically formed by a set of spatially confining localities,
intended as the “solutions” in which chemical interactions
occur and across which chemicals can eventually diffuse.
The declarative tuple space model of TuCSoN coordination
infrastructure can support an effective implementation for
chemically-inspired interactions [
In the case study, we can think of display services, of information services (concerning ads and news), as well as of user and environmental data as molecules. Displays represent different localities (i.e., tuple spaces) in which components react. Chemical rules dictate that when the preferences of a user entering the locality of a display match an information/advertisement service, then a new composite component is created which is in charge of actually displaying that service in that display. Concurrently, in each locality, catalytic components can be in charge of re-enforcing the concentration of specific information or of specific information/advertisements, reflecting the current situation of users. Also, localities can be open to enforce chemical bonds across displays, so that highactivity of advertisement reactions on a display can eventually propagate to neighbors.
Chemical metaphors can effectively lead to self-organizing
structures like local composite services and local aggregates.
As in real chemistry, chemical computational metaphors can
accommodate an incredible amount of different components
and composites with a single set of basic laws. In practice,
this means that it can tolerate an increasingly diverse and
evolving set of semantic descriptions for components without
affecting basic eco-laws and without increasing the burden to
the infrastructure. As far as control is concerned, one can think
at using sort of catalyst or reagent components to engineer
and control (in a fully decentralized way) the dynamics and
the behavior of the ecosystem. A limitation of the chemical
approach is that it typically relies on activities taking place
within a locality or at least across neighboring ones via local
diffusion [
These focus on biological systems at the scale of individual organisms, or of colonies of organisms like ants. The species are therefore either simple cells or animals acting on the basis of simple goals like finding food and reproducing (to be modeled and implemented in terms of simple goal-oriented agents). As in physical systems, interactions take place by means of signals of various flavors (i.e., chemical pheromones) spread by individuals in the environment, and slowly diffusing and evaporating. The spatial environment is again a computational landscape either mapped on the network topology or on the physical space. Unlike physical systems, individuals here are not necessarily passively subject to the sensed pheromones, but they can react to them depending on their current “mood” (e.g., their state towards the achievement of a goal). Consequently, eco-laws are only aimed at determining how such pheromones, depending on their specific flavors, should propagate and diffuse in the environment. From the implementation viewpoint, any infrastructure that can support physical metaphors can also be adapted to support biological metaphors, by turning fields into persistent and slowly diffusing/evaporating data structures.
In the case study, users will be represented by simple agents roaming around and spreading chemical signals with a flavor reflecting their personal interests. Displays can locally perceive such pheromones, and react by emitting some different pheromones to express the availability of commercials and information. Advertising and information agents, by their side, sense the concentration of such pheromones and are attracted towards the displays where the concentration of users with specific interests is maximized. Displays, advertising and information agents by their side, and depending on what they have displayed so far, can also emit additional flavors of pheromones, to store memory of past events. The persistence of pheromones can also be exploited by additional components that, by moving from display to display, create pheromones trails as a basis for more global strategies—like identifying routing paths that advertiser agents use to globally find the best displays to exploit.
Biological metaphors appear very flexible in efficiently
enabling the spatial formation of both localized and distributed
activity patterns, and have a variety of applications [
These focus on biological systems at the level of animal
species and of their interactions [
In the case study, and assuming that each display is associated to an ecological niche, we can imagine users as species of herbivor agents roaming from niche to niche, possibly to eat those vegetable-like individuals representing information. Advertisers can be sorts of carnivors (i.e., eating other active life-forms) in need to find proper users to survive (users with matching interests). For both cases, the primary effect of an eating action is the feeding of displays with information or commercials to show. A possible secondary effect of eating is the reproduction and diffusion in the environment of the best-fit species, e.g., of most successful commercials. Those species that do not succeed in eating at a niche can either die or move to other niches to find food. Concurrently with the activities of the above species, background agents acting as sorts of bacteria can digest the activity logs at the various niches to enforce specific forms of distributed control over the whole system (e.g., by affecting the way information is propagated across niches).
Social metaphors, such as the one here described but without forgetting the large body of work in the area of social multi-agent systems, appear suitable for local forms of selforganization like chemical metaphors—think at self-organized equilibria of web food patterns in ecological niches. Unlike chemical metaphors, and more similarly to biological one, social metaphors can be effectively exploited also to enforce distributed forms of self-organization, by exploiting the capability of individuals of moving around (e.g., to find food and/or reproduce). In addition, social metaphors are suitable for tolerating diversity and evolution at no additional burden for the infrastructure—think at how biodiversity has increased over the course of evolution without requiring any change to the underlying infrastructure (i.e., the earth). However, understanding how to control the behavior and dynamics of social computational ecosystems can be as difficult as it is in real social systems.
Nature-inspired models and architectures appear very promising for re-thinking the foundation of modern and emerging pervasive service systems. Yet, the general lesson that we can extract from our analysis is that:
None of the natural metaphors being proposed so far is suited, as is, to act as a way to fully realize the eternally adaptive service ecosystem vision.
Thus, along the lines of the proposed reference architecture and being guided by it, some new synthesis should be identified to incorporate the key features of the existing metaphors into a unifying general-purpose one.
In addition to the key problem of identifying suitable
metaphors, the widespread deployment of nature-inspired
pervasive service ecosystems requires methodologies and tools
for their engineering, proper security mechanisms and policies,
and means to integrate the approach with the legacy of current
SOA systems. Lastly, the road towards pervasive service
ecosystems will have to ground on the emerging science of
service systems [