A CONTEXT MODEL FOR AUTONOMIC MANAGEMENT

OF AD-HOC NETWORKS

Anastasios Zafeiropoulos, Athanassios Liakopoulos

Greek Research & Technology Network, Av. Mesogion 56, 11527, Athens, Greece

Panagiotis Gouvas

Ubitech, Av. Mesogion 429, 15343, Athens, Greece

Keywords: Context model, Ad-hoc network, Autonomic management, Policies, Distributed repository.

Abstract: Management of next generation networks is challenging due to increased complexity imposed by their

dynamic and heterogeneous characteristics. The deployment of mesh wireless networking topologies, the

support of diverse networking functionalities and the existence of large number of heterogeneous devices

make traditional approaches inappropriate. In such environments, the description of the basic networking

entities and the interactions that are present, as well as the relationships among them, is crucial. Proper

representation may facilitate the operation and management of the network, as well as the optimal

adaptation to the current environmental conditions, and thus, optimise the performance of the network

mechanisms. In this paper, a context model is proposed for ad-hoc networks aiming to present in detail the

correlation among the network entities and interactions in dynamic environments. Specific functionalities

that may be designed taking in account the description of the context model are described and indicative

implementation scenarios are implemented and evaluated.

1 INTRODUCTION

Next generation networks are becoming more

complex and dynamic through the interconnection of

large number of heterogeneous devices, the

deployment of mesh networking topologies and the

support of diverse networking functionalities.

Ad-hoc communication among independent nodes in

a network, frequent topology changes due to

inherent dynamic network characteristics, exchange

of roles among the nodes in a network and merging

or splitting of groups of nodes are expected to be

dominant characteristics of future networking

enviroments.

The Future Internet is envinsioned to become a

service-aware networking environment where each

node will be able to undertake the role of the service

or content provider. Services may be composed or

decomposed according to the availability of

resources and applied policies while, in addition,

provision of services have to be conducted in a

unified way.

By taking into account this evolution, there is a

necessity to design new paradigms to accomplish

efficient management of next generation networks.

The incorporation and combination of concepts like

self-adaptation, self-configuration, self-awareness

and self-healing to the current networking conditions

as well as the support of decentralized management

schemes is required. In this manner, high level goals

in the network may be achieved in an autonomous

manner without the need for predefined centralized

infrastructure.

A promising method for the distributed and self-

management of next generation networks, as well as

the provision of advanced services in dynamic

environments, is the establishment of logical

overlays on top of existing physical networks. In an

overlay network, knowledge may be shared among

the network nodes and the network may be

re-organized according to the nodes’ content or the

supported functionalities. The underlying network

topology changes may be hidden and reliability in

the provision of services may be assured.

Zafeiropoulos A., Liakopoulos A. and Gouvas P. (2011).

A CONTEXT MODEL FOR AUTONOMIC MANAGEMENT OF AD-HOC NETWORKS .

In Proceedings of the 1st International Conference on Pervasive and Embedded Computing and Communication Systems, pages 74-82

DOI: 10.5220/0003368900740082

 SciTePress

In addition to the establishment of overlay

networks, knowledge representation and proper

interpretation of the acquired context is crucial for

the support of complex management tasks. Through

appropriate context representation, each node in the

network may be able to efficiently sense its

environment through the establishment of interaction

with neighbouring nodes and extract knowledge

based on it (self-awareness). Moreover, concepts

like the quality of collected context (QoC)

(Buchholz, 2003) (Krause, 2005), the supported

functionalities and roles of each node, the scope of

the collected information (a.k.a. whether the

information is useful for neighbouring nodes or for

all the nodes in the network) and the node and

network temporal characteristics (e.g. mobility ratio,

location etc.) may be efficiently expressed and

combined for information fusion purposes. Ad-hoc

nodes may dynamically evaluate the acquired

context according to predefined rules and exchange

knowledge based on it, while knowledge of the

existing policies and conditions in the network may

facilitate the distributed decision making process.

Through proper allocation of roles within the

network, distributed reasoning may also be achieved

in local, cluster and network level.

In this paper, focus is given on the description of

a context model that describes the basic entities and

interactions that are present in dynamic networking

environments. The designed context model

expresses and correlates the networking entities in

an ad-hoc network, their environmental

characteristics and the imposed policies on the

network, aiming at facilitating the operation and

management of the network, as well as optimising

the performance of the network mechanisms. The

presented approach extends our previous work on

designing an architecture for autonomic services

provision in dynamic networks through the

establishment and maintenance of an overlay

network (Gouvas, 2010). An overlay network may

be used as distributed repository of context aware

information that can be shared among the participant

ad-hoc network nodes.

The paper is organized as follows; Section two

briefly presents the current work in the field of

context awareness support in dynamic

heterogeneous networks. Section three describes the

requirements that have to be fulfilled for the proper

design of a context model for ad-hoc networks,

details the proposed context model, as well as the

basic functionalities that may be supported. Section

four describes the emulation results based on the

optimisation of specific network mechanisms and,

finally, Section five concludes the paper with a short

summary of our work and a discussion of open

issues and future work.

2 RELATED WORK

Several approaches have been proposed for

autonomic management of ad-hoc networks. These

include techniques for distribution of functionalities

and roles in a dynamic environment, design of

context models for description of interactions within

an ad-hoc network as well as methodologies for

distributed processing and reasoning of context data.

The design and implementation of a system that

exploits context-awareness and couples it with

policy-based management in order to enable the self-

management of mobile ad-hoc networks (MANETs)

is presented in (Malatras, 2007). It is proposed that

management of MANETs has to be done in a

hierarchical but also distributed manner through a

dynamically constructed set of manager nodes.

Similarly, the vision of realizing the idea of a “self-

organized data ecosystem” by defining proper

models and tools to represent, analyze, self-organize,

and self-aggregate contextual information, so as to

form structured and meaningful collections of

related knowledge items is described in (Bicocchi,

2010). It is stated that the evolution must be from

models of context awareness in which the focus is to

provide services with simple interfaces to access

heterogeneous context providers, towards models of

situation awareness in which a middle layer is in

charge of organizing sparse pieces of information in

order to provide services with a pre-digested and

more comprehensive higher level knowledge.

A layered reference model that encapsulates

suitable abstractions to tackle the complexity of

context management in mobile ad-hoc networks is

proposed in (Dudkowski, 2008). The proposed

model comprises of three major layers, each of

which addresses a specific part of the overall

complexity of context management: core storage,

update processing, and context services.

Respectively, in (Christopoulou, 2010), focus is

given on the exploitation of context in ubiquitous

computing environments through its modelling using

ontologies and on the reasoning process based on

intelligent context and rules defined by the user.

Important effort is also dedicated on the

description of techniques for distributed data and

knowledge acquisition. Context data distribution is

defined as the capability to gather and timely deliver

relevant context data about the environment to all

A CONTEXT MODEL FOR AUTONOMIC MANAGEMENT OF AD-HOC NETWORKS

interested entities in an ad-hoc network (van

Sinderen, 2006). A simple way to specify context

data distribution requirements is proposed in

(Corradi, 2010), along with an approach for quality-

aware context data distribution in mobile and

wireless wide-area environments. In (Macedo, 2009)

a distributed entity and an information schema is

described to store and disseminate information

concerning the network, its services and the

environment, orchestrating the collaboration among

cross-layer protocols, autonomic management

solutions and context-aware services. Moreover, in

(Rizou, 2010) the problem of designing a distributed

system for the detection of situations in highly

dynamic environments is addressed. An architecture

is introduced that enables the efficient distributed

processing of context data in large-scale overlay

networks.

2.1 Motivation & Contribution

Based on the already proposed approaches, it may be

argued that through the support of context awareness

in dynamic networking environments, self-

adaptability can be achieved in the supported

mechanisms and network nodes may fulfil network-

level goals that derive from high-level policies.

However, the description of a representative context

model for interactions in ad-hoc networks that

combines and presents the concepts of dynamicity,

distribution of roles and resources, knowledge

extraction through interaction of the ad-hoc network

nodes and adaptation to the existing mechanisms in

the network according to the imposed goals and

policies, need to be further investigated.

Furthermore, techniques that enable the distributed

storage and dissemination of knowledge in the

network are not analyzed in detail in the literature.

The motivation for the design of the context

model that is proposed in this paper is the

description of a reference model for the

implementation of context management in MANETs

that combines the above mentioned concepts. The

proposed context model is generic enough in order

to support diverse networking environments,

functionalities and deployment scenarios. The

description of the context model is based on an

already proposed architecture for providing

autonomous and decentralized services in dynamic

environments through the establishment and

maintenance of overlay networks, as it is described

in (Gouvas, 2010). Overlay networks may be used as

distributed repositories where context aware

information may be shared among the participant ad-

hoc network nodes and service development may be

realized independently from the underlying physical

network. Finally, mechanisms that may be applied in

large scale networks (e.g. clustering mechanisms in

wireless sensor networks (Zafeiropoulos, 2010)) are

also considered.

3 CONTEXT MODEL FOR

AD-HOC NETWORKS

3.1 Requirements

An efficient context model for ad-hoc networks

should fulfil the following requirements (Malatras,

2007): a) be extensible in order to describe a rich set

of interactions, concepts and functionalities, b) be

interoperable with existing models, c) support

accurate descriptions and d) require limited memory

and processing power in order to be suitable for

resource constrained devices.

Proper representation and description of the

context information is important. Context

information is considered as incorrect if it fails to

reflect the true state of the world it models. It is also

regarded as inconsistent, if contains contradictory

information. Also, it is considered as incomplete if

some aspects of context are unknown (Henricksen,

2002). Thus, in order to accurately describe the

acquired context information, it is desirable to

achieve high Quality of Context (QoC) (Buchholz,

2003) (Krause, 2005). Various parameters may be

used to characterize the quality of (context)

information from different perspectives, such as

freshness (i.e. time passed since the creation of the

content), precision in the estimation, scope (e.g.

local, cluster or network level), trust-worthiness,

privacy, retrieval time from the content requestor,

replication degree, volatility, etc.

The context model should fulfil several

requirements imposed from the dynamicity of ad-

hoc networks. Different roles have to be supported

for the network nodes (e.g. router, gateway, cluster

head) since delegation of roles is a dynamic and

evolving process and depends on the nodes

capabilities and the overall network policies.

Moreover, concepts like mobility (e.g. frequency of

link breaks with neighbours), available energy and

network topology formation (e.g. size and degree of

the network) have to be expressed. The definition of

a hierarchical structure that the ad-hoc network

nodes may follow as well as the possible ways of

interaction among the network nodes during the

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems

information flow procedure have also to be

described. The scope for each type of data (e.g.

local, cluster, global level) has to be defined and be

interconnected with the type of the repository that

will be used (e.g. distributed repository, centralized

database).

Finally, the context model has to support

description of policies and available services in the

network. The definition of a policy has to be in a

suitable format, able to be correlated with the

represented context data. Policies definition has to

be conducted according to the high level goals that

are imposed by the network administrator and the

events that are extracted from the context model and

the services description.

3.2 Context Model Description

The proposed context model (Figure 1 – Figure 4)

aims at describing the existing entities and

interactions in an ad-hoc network as well as basic

functionalities that these entities may support. The

basic entities of the context model are the Node that

is able to proceed to Decisions, to undertake Actions

and to conduct proper Reasoning over the acquired

context. The Node supports specific Functionalities

(e.g routing, security, QoS protocols and

mechanisms, etc) and, through a decision making

process, is able to affect their Configuration

Parameters. This is necessary in order to support

self-optimisation and self-adaptation to the existing

conditions in the network. Furthermore, according to

the supported Functionalities, each Node may be

associated with specific Roles (e.g. cluster head in

case of application of clustering mechanisms)

(Figure 1).

The Node contains Monitoring/Sensory

Interfaces through which it senses specific Context

Parameters. Each Monitoring/Sensory Interface is

able to measure or sense networking parameters (e.g.

size and density of the network), mobility

parameters (e.g. percentage of dynamicity in the

network), connectivity parameters (e.g. list of

neighbours), environmental metrics (e.g. sensor

measurements) and the node status (e.g. available

energy). Each Context Parameter has specific

Parameter Attributes that express attributes related

with the QoC for this parameter. These attributes

include freshness, scope (local or global), volatility,

privacy, replication degree, confidence interval,

sampling frequency and user-related community,

etc. According to the scope of the acquired data, it

may be stored either at the Distributed Repository or

-centrally or locally- in specific Nodes (Figure 1).

Figure 1: Supported functionalities and roles by each

network node.

Figure 2: Node attributes and user profiles.

The Node has Node Attributes and Network

Attributes that may be constant or temporal. Node

Attributes include the available battery, storage and

memory, the CPU, the maximum forwarding

capacity and a Key Performance Indicator (KPI) that

may be variable according to each application needs

(e.g. mainly computed based on available energy on

power-constrained nodes). Network Attributes

include connectivity attributes such as the current

location and the list of neighbours, mobility

attributes such as the number of link breaks with

neighbours in a specific time period, and

environmental metrics. The Node has also a global

identifier (e.g. IPv6 address), specific

communication capabilities and supports specific

User Profiles (Figure 2).

An important part of the model is the policy-

based decision making process description. The

description is based on the aim of fulfilment of the

specified Goals in the network that may be related

A CONTEXT MODEL FOR AUTONOMIC MANAGEMENT OF AD-HOC NETWORKS

with the technical or business perspective. These

Goals are described through Policies based on the

selected policy description language (e.g. Common

Information Model Simplified Policy Language

CIM-SPL (DMTF CIM-SPL, 2010)). Each Policy

has a specific applicability scope since it may regard

a local or network wide decision and is comprised of

Rules. The Rules are based on combination of

Conditions and Events that are even described

including specific Context Parameters or combine

also Knowledge that is extracted based on the

Node’s Reasoning process. The advantage of this

definition is that Knowledge that is produced during

the network operation and evolution may be

exploited and specific Events and Context-aware

Parameters may be also used in the Rules definition

process. The Rules have applicability and storage

scope and may be stored either in the Distributed

Repository or -centrally or locally- in specific

Nodes. Depending on the network environment and

the supported services characteristics, Rules may be

preferable to be commonly accessible for all

network nodes including those that join currently the

network. In this case, their storage in a Distributed

Repository is recommended. Rules that reason on

decisions and actions based on global knowledge

may be similar with Rules that are based on local

knowledge. According to their applicability state,

they will be applied in the appropriate region (Figure

3).

Figure 3: Network goals, policies and rules.

The imposed Policies are also associated with

the supported Services in the network that have

specific QoS characteristics and priorities. Each

Service may initiate several Flows in the network

that are established within end-to-end paths. The

Flows are characterized based on the source node,

the destination node, their priority and QoS

characteristics, the signalling information and maybe

from a flow record (e.g. as defined in IPFIX). The

context model is designed in order to be compatible

with existing models for description of policies,

services, monitoring and QoS characteristics (DMTF

CIM, 2010) (TMF, 2005) (Schmohl, 2008) (Ferreiro,

2009).

Another important aspect in the design of the

context model is the specification of the way that

information may be disseminated in the ad-hoc

network. This information regards the available data

in the network as well as the defined policies and

rules. As stated earlier, information may be stored

either in specific Nodes (locally or centrally) or in

the Distributed Repository. Information exchange

may be accomplished via interrupt based or polling

based techniques. In order to succeed distributed

storage and dissemination of information, the

approach proposed in (Gouvas, 2010) may be

followed or alternative methods and techniques may

be designed and developed.

Figure 4: Dissemination of information within the

network.

3.3 Supported Functionalities

Several functionalities may be designed according to

the specified context model for improving the

operation and management of the ad-hoc network.

Indicatively, roles may be assigned in the network

nodes according to their capabilities based on the

definition of a Key Performance Indicator (KPI)

function. Similarly, according to the dynamicity that

is present in the network several actions may be

taken, such as a) increase in the data replication

degree in the overlay network in order to ensure that

valuable data will not be lost and b) re-initiation of

mechanisms for autonomic estimation of network

based parameters for acquiring an up-to-date view.

The percentage of dynamicity in the network may be

given by the ratio

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems

mobility_ratio=link_breaks/neighbours (Buchholz,

2003) for a specific time period. Adaptations may

also be supported in the routing protocols applied in

the ad-hoc network. According to the levels of

dynamicity in the network, the use of proactive or

reactive routing protocols may be preferable (e.g. in

case of a highly dynamic environment, reactive

routing protocols have to be selected).

Furthermore, modifications and self-optimisation

actions may be supported in mechanisms that are

mostly applied in large scale networks (e.g. large

wireless sensor networks). An important

functionality in such networks is usually the

aggregation of data by special nodes and the

adaptation in the frequency of collecting data

according to the network conditions. Responsible

nodes for aggregation are defined taking in account

also the context scope. Threshold criteria may be

defined for specific actions according to the

collected data (e.g. when the available energy in the

cluster is low, reduce the frequency of data

sampling). Similarly, in case of application of

clustering mechanisms, if the cluster head (CH)

senses reduction in its available energy, it may quit

from this role and return to normal operation.

The context model may be used in various

network monitoring and management scenarios,

since interoperability with existing context models

related with traffic monitoring and QoS provision is

supported. Through interaction with the imposed

policies and the supported rules in the network, self-

optimisation and self-configuration decisions may be

taken. An indicative scenario for autonomic traffic

monitoring and management, based on the designed

context model is described in section 4.

The context model also supports information

regarding user profiling and preferences. This

information is usually proactive (known

beforehand). According to this information, the user

may be correlated with existing content in the

network (e.g. content characterized for this

community). Finally, security and privacy issues

declared by the user may impose specific

adaptations in the functionality of the supported

mechanisms.

4 EXPERIMENTAL RESULTS

& EVALUATION

Two indicative scenarios are examined in this

section, in which the identified interactions are

described according to the proposed context model.

Based on tests results, we claim that significant

improvements may be achieved by dynamically

adapting a routing protocol or an application-level

mechanism in an ad-hoc network. Focus, though, is

given on the applicability of the context model in

diverse networking environments rather than on

optimising the selected protocol or mechanism.

The experiments are conducted in an emulation

environment where a wide set of topologies are

formed by initiating multiple mobile nodes. The

prototype implementation is developed in Java and

supports the bootstrapping of a (multi-hop) ad-hoc

network and the communication among the

participating nodes. A topology editor is also

implemented in order to examine the various

mechanisms in specific topologies.

Table 1: Description of scenario 1 according to the context

model.

<condition-set operator="and">

<operant name="Network_size"

scope=”global” value="N"/>

<operant name="Network_density"

scope=”global” value="d"/>

</condition-set>

<action-set>

<config layer="Routing"

param_name="Route_Req_HTL"

param_value="f(N,d)" />

<config layer="Routing"

param_name="Route_Error_HTL"

param_value="f(N,d)" />

</action>

</action-set>

</rule>

</ruleset>

</goal>

In the first scenario, a network with diverse size

and density bootstraps. It is assumed that the

network nodes are able to estimate the size and the

density of the network through the application of

gossiping techniques for estimation of network

based parameters (Gouvas-MONET, 2010). A

Constant Bit Rate (CBR) flow may be established

between any two nodes in the network and the

routing performance is estimated for the following

cases: i) the Hops-To-Live (HTL) parameter is

dynamically adapted according to the current size

and density of the network, and ii) the Hops-To-Live

(HTL) parameter is set to infinite.

A CONTEXT MODEL FOR AUTONOMIC MANAGEMENT OF AD-HOC NETWORKS

In Table 1, the description of the Goals, the

Rules and the Actions for the realization of the

scenario is shown based on our proposed context

model. The Goal for the network is to reduce the

unnecessary flooding of messages. A Rule is defined

for specific Actions according to the values of the

size (N) and density (d) global parameters. The HTL

value for routing request (Route_Req_HTL) and

routing error messages (Route_Error_HTL) is

defined according to a function f(N,d). Thus, during

the network bootstrap or after a network topology

change, the Rule is executed and the HTL parameter

value is adapted accordingly. The total number of

messages exchanged in mesh and Barabasi-Albert

topologies is shown in Figure 5, where the size and

density of the network vary from 10 to 40 nodes and

from 2 to 4, accordingly.

Figure 5: Total number of messages exchanged in mesh

and Albert-Barabasi topologies.

In the second scenario, an Albert-Barabasi

network topology is established for various network

size and density. At specific time periods, traffic

flows are initiated among network nodes and a

specified sink node. There are two types of flows;

CBR flows with rate of 10Mbps, and monitoring

flows with rate of 1Mbps. For each CBR traffic

flow, a monitoring flow is initiated between the

same source and the destination node for

management purposes. The monitoring flow is able

to collect measurements and statistics for specific

traffic performance metrics, e.g. delay, jitter,

congestion, etc. According to the collected

measurements, distributed monitoring and

management may be performed in the ad-hoc

network, as it is described in (Liakopoulos, 2010).

The Goal in the network is to achieve high

monitoring accuracy without posing negative impact

in the generated traffic. According to the existing

policy, the monitoring accuracy is desired to be as

high as possible provided that there is no congestion.

In case of congestion, the monitoring accuracy

should be reduced to a minimum level in order to

free scarce resources. Therefore, in cases that the

generated traffic is increased significantly and

congestion may appear in the network, traffic

generated by monitoring flows is eliminated

(throttling). The following Rule is imposed for this

purpose: when a node serves traffic more than

30 Mbps, it drops the signalling (monitoring)

packets. If the served traffic goes again under

30 Mbps in the future, signalling packets are

forwarded. In Table 2, the description of the Goals,

the Rules and the Actions for the realization of the

scenario is shown based on the designed context

model.

Table 2: Description of scenario 2 according to the context

model.

<ruleset

name="throttle_Monitoring_Pkts">

<condition-set operator="and">

<operant name="link_Utilisation"

scope=”local” value="N"/>

</condition-set>

<config layer="Application"

param_name="Monitor_Pkts_Served"

scope="local"

param_value="f(N)" />

</action>

</actionset>

</rule>

</ruleset>

</goal>

Figure 6: Monitoring flow packets served.

The total number of messages for the monitoring

flows and the CBR traffic flows in case where the

traffic monitoring Rule is applied or not is depicted

in Figure 6 and Figure 7, accordingly. It is shown

that, in case of application of the defined Rule, the

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems

number of served monitoring packets in the network

is reduced in case of high utilisation in comparison

with the case where no rule is imposed. Similarly,

the total number of messages served in the case

where the Rule is not applied is larger. However, in

case of congestion, the QoS provision could be

deteriorated in this case compared with the case

where the Rule is applied.

Figure 7: CBR flow packets served.

5 CONCLUSIONS

In this paper, a context model for ad-hoc networks is

proposed. The model takes in account the need for

proper representation of the network entities and

their interactions that are present in various dynamic

environments. It also aims to improve the context

awareness level in the network and, thus, facilitate

the realisation of autonomic management

mechanisms. In addition, the dynamic adaptation of

protocols to the current network conditions may also

enable the realisation of self-optimised functions

among independent nodes.

In our future work, we plan to design more

complex scenarios based on the proposed context

model and present the optimisation that may be

succeeded by allowing dynamic adaptation of

network protocols and mechanisms. Furthermore, an

important research issue is related with the design

and development of methods for providing

distributed reasoning functionalities in a dynamic

environment. Novel techniques for rules and policies

storage and distribution in the network, as well as

intelligent ways for autonomic and distributed

decision making mechanisms have to be further

examined.

ACKNOWLEDGEMENTS

This publication is based on work partially

performed in the framework of the European

Commission ICT/FP7 project EFIPSANS

(www.efipsans.org).

REFERENCES

Buchholz, T., Kpper, A., Schiffers, M., 2003. Quality of

context: What it is and why we need it. In Proceedings

of the 10th International Workshop of the HP

OpenView University Association (HPOVUA),

Hewlet-Packard OpenView University Association.

Krause, M., Hochstatter, I., 2005. Challenges in modeling

and using Quality of Context (QoC). Mobility Aware

Technologies and Applications, LNCS, vol. 3744, pp.

324–333, Springer Berlin / Heidelberg.

Gouvas, P., Zafeiropoulos, A., Liakopoulos, A., Mentzas,

G., Mitrou, N., 2010. Integrating Overlay Protocols for

Providing Autonomic Services in Mobile Ad-hoc

Networks, IEICE Communications, Special Issue on:

Implementation, Experiments, and Practice for Ad

Hoc and Mesh Networks, vol. E93-B, no. 8.

Malatras, A., Hadjiantonis, A. M., Pavlou, G., 2007.

Exploiting Context-Awareness for the Autonomic

Management of Mobile Ad Hoc Networks, Journal of

Network and Systems Management, vol. 15, pp. 29-

55, Springer New York.

Bicocchi, N., Baumgarten, M., Brgulja, N., Kusber, R.,

Mamei, M., Mulvenna, M., Zambonelli, F., 2010. Self-

Organized Data Ecologies for Pervasive Situation-

Aware Services: The Knowledge Networks Approach,

IEEE Transactions on Systems, Man and Cybernetics,

Part A: Systems and Humans, vol.40, no.4, pp.789-

802.

Dudkowski, D., Weinschrott, H., Marron, P.J., 2008.

Design and Implementation of a Reference Model for

Context Management in Mobile Ad-Hoc Networks,

22nd International Conference on Advanced

Information Networking and Applications -

Workshops, pp.832-837.

Christopoulou, E., Garofalakis, J., 2010. Enabling the

User to Setup Ubiquitous Computing Applications

Based on Intelligent Context, International Conference

on Complex, Intelligent and Software Intensive

Systems (CISIS), pp.87-92.

van Sinderen, M. J., van Halteren, A. T., Wegdam, M.,

Meeuwissen, H. B., Eertink, E. H., 2006. Supporting

context-aware mobile applications: an infrastructure

approach, IEEE Communications Magazine, vol.44,

no.9, pp.96-104.

Corradi, A., Fanelli, M., Foschini, L., 2010. Adaptive

context data distribution with guaranteed quality for

mobile environments, 5th IEEE International

Symposium on Wireless Pervasive Computing

(ISWPC), pp.373-380.

A CONTEXT MODEL FOR AUTONOMIC MANAGEMENT OF AD-HOC NETWORKS

Macedo, D., Dos Santos, A., Nogueira, J. M. S., Pujolle,

G., 2009. A distributed information repository for

autonomic context-aware MANETs, IEEE

Transactions on Network and Service Management,

vol. 6, no. 1, pp. 45-55.

Rizou, S., Haussermann, K., Durr, F., Cipriani, N.,

Rothermel, K., 2010, A System for Distributed

Context Reasoning, Sixth International Conference on

Autonomic and Autonomous Systems (ICAS), pp.84-

89.

Zafeiropoulos, A., Gouvas, P., Liakopoulos, A., Mentzas,

G., Mitrou, N., 2010. NEURON: Enabling

Autonomicity in Wireless Sensor Networks, Sensors

10, no. 5, pp. 5233-5262.

Henricksen, K., Indulska, J., Rakotonirainy, A., 2002.

Modeling Context Information in Pervasive

Computing Systems. In Proceedings of the First

International Conference on Pervasive Computing

(Pervasive '02), Friedemann Mattern and Mahmoud

Naghshineh (Eds.). Springer-Verlag, London, UK,

167-180.

Liakopoulos, A., Zafeiropoulos, A., Marinos, C.,

Grammatikou, M., Tcholtchev, N., Gouvas, P., 2010.

Applying distributed monitoring techniques in

autonomic networks, 2nd IEEE International

Workshop on Management of Emerging Networks and

Services (IEEE MENS 2010), in conjunction with

IEEE GLOBECOM 2010, Miami, Florida, USA.

DMTF, CIM Simplified Policy Language (CIM-SPL),

Available Online: http://www.dmtf.org/sites/

default/files/standards/documents/DSP0231_1.0.0.pdf

(last visited November 2010).

DMTF, Common Information Model (CIM) Standards,

Available Online: http://www.dmtf.org/standards/cim/

(last visited November 2010).

TMF, Shared Information/Data (SID) Model - Business

View Concepts, Principles, and Domains. GB922, Ed.

NGOSS R6.1, Document Version 6.1, November,

2005.

Schmohl, R., Baumgarten, U., 2008. A Generalized

Context-aware Architecture in Heterogeneous Mobile

Computing Environments, The Fourth International

Conference on Wireless and Mobile Communications,

pp.118-124, Washington, DC.

Ferreiro, A., Fichtel, T., Lopez de Vergara, J., Matray, P.,

Strohmeier, F., Tropea, G., Weinsberg, U., 2009,

Semantic Unified Access to Traffic Measurement

Systems for Internet Monitoring Service, ICT-Mobile

Summit 2009, Santander, Spain.

Gouvas, P., Zafeiropoulos, A., Liakopoulos, 2010.

Gossiping for Autonomic Estimation of Network-

based Parameters in Dynamic Environments, Fifth

International Workshop on MObile and NEtworking

Technologies for social applications (MONET´10), in

conjunction with OnTheMove 2010

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems