An E-learning Service Management Architecture

An E-learning Service Management Architecture

Theodore K. Apostolopoulos

Anna Kefala

Department of Informatics

Athens University of Economics and Business

76 Patission st., 10434 Athens, Greece

Tel.: +30 210 8203234, 163 Fax: +30 210 8203159

Email: {thodoros,anna}@aueb.gr

Abstract

The aim of this paper is proposing a configurable architecture for implementing e-learning service management, taking into account several diverse aspects concerning not only the educational process, but also network infrastructure, the QoS requirements, the types of educational content, etc. We present a flexible, parametrical and component-oriented management scheme, which dynamically creates a session profile for each request for e-learning service according to current context that depends on all diverse aspects that an e-learning service comprises. Continuously, we analyse the components comprising the management scheme and the management information bases that they maintain.

1.

Introduction

Today’s picture of computer networks area is composed by a very heterogeneous environment supporting multivendor applications upon a variety of underlying switching systems and transmission facilities [11]. The need to control this continuously growing in scale, complex and heterogeneous environment, has triggered an emerging need for managing these networks. According to latest technological trends, however, traditional computer data networks are evolving into a broader communication means (telecommunication networks), which will provide infrastructure not only for simple data transportation but also for sophisticated electronic services.

The management of telecommunication networks mainly aims at providing a reliable and efficient infrastructure and means of transporting information among distributed and diverse applications and services. Nevertheless, when considering a network-based service, we cannot easily discriminate between the application and the underlying network infrastructure and isolate each one. Every day new applications and services deploy telecommunication infrastructure as a vehicle to reach new clients, located in

different and distant areas. The need for providing distributed services, leads to an emerging need for managing these services as a whole, including the management of underlying network infrastructure.

During the last years a rapidly increasing number of educational and research institutions along with companies and other organizations adopt a new educational paradigm, that of a distance education or electronic learning (e-learning), as their emerging choice and begin to offer e-learning services. E-learning is primarily characterised by physical distance between instructor and learner and possibility of group or individual synchronous or asynchronous participation in the educational process. Its concept includes an emulation of the educational environment in terms of roles (student, tutor or facilitator) undertaken and educational methods. The evolvement of different communication technologies offers the possibility of delivering a variety of media, from printable material to live satellite videoconferences and electronic messages to the Internet.

Recently many efforts have been made to build online learning systems, which manage the e-learning process. Such systems are often named Learning Management Systems (LMS). However, many centralized LMSs are not flexible enough to deal with the personalized and dynamic learning needs of individual learners, [4] when more sophisticated LMSs which have the ability to dynamically adapt the material and to present instructional modules according to the learner’s needs still do not take into account other parameters that might vary per each e-learning session and mostly focus on managing content and registration related issues.

Therefore, the need for an e-learning management framework that would combine several requirements and enable communication and cooperation between some e-learning software and the underlying network infrastructure is emerging [1]. Such a framework could comprise an intermediate layer which would concentrate all required information and perform the management of

the whole e-learning service, including the communication between user and the e-learning server or the communication among distributed e-learning servers. In the following section, a review of the proposed e-learning service framework is presented. In section 3, a detailed description of the proposed e-learning service management architecture is presented, while in section 4, the managed components are analytically described. We conclude with some ideas about future work to be done.

2.

E-learning service framework

When designing and implementing a distance learning system or an e-learning service in general, several aspects are taken into consideration. These aspects concern not only the educational and pedagogical process or the educational content, but also the type of the offered service (e.g. synchronous, asynchronous or self-paced), the network and physical layer infrastructure, the user needs, the functionalities of the software used for the implementation of the service, etc.

Most times, during the design and implementation of e-learning systems, only some specific aspects and factors are supported, resulting in systems that offer partial services. Therefore we have noted the lack of a middleware that would bridge the gap among the several aspects forming an e-learning service and would consist a balanced solution based on conflicting requirements [1]. A middleware that would implement a consistent management scheme for the whole e-learning service. In the proposed middleware architecture [1], we consider an intermediate layer between an e-learning application and the underlying network infrastructure. Beyond it’s role as a facilitator for the cooperation and communication between upper and lower layers, middleware will be responsible for gathering all required information from the components comprising an e-learning system in order to manage the whole service, as presented in Figure 1.

Figure 1. Proposed Middleware Architecture The upper layer of proposed architecture is occupied by some e-learning application. There are no special restrictions regarding the characteristics of such an application. It will communicate with the middleware through specific Application Program Interfaces (APIs).

When discussing about network infrastructure, two separate issues are on focus:

ÖPhysical Network Infrastructure

This term refers to diverse physical network technologies available for telecommunication and tele-cooperation purposes.

ÖNetwork-Communication Protocols

These protocols correspond to network and upper layers (in correspondence to OSI model architecture). As far as the network layer protocol is concerned, Internet Protocol (IP) is the dominator. As far as upper layers protocols are concerned, diverse protocols are available depending on application’s requirements and characteristics of information to be transferred.

Middleware platforms traditionally have black box architecture [6]; in our approach, we open up this black box and encourage flexibility and adaptability by choosing a component-oriented architecture that would be configurable. The proposed management scheme will be composed by several components in order to achieve communication with different types of e-learning applications, while at the same time will cooperate with different plug-ins depending on the available network infrastructure. The component oriented design approach allows the design of learning systems that are context-independent; while the focus on autonomous components allows a complex system to be considered as an arbitrary number of smaller cohesive components [4].

The proposed middleware aims at configuring in real-time the e-learning sessions by creating them in a dynamic and flexible way. Due to the number of potential users to support, we need to adopt a decentralized system with specific components or agents, handling the communication between distributed participants. An agent manages individual and collective learning, adapted to the specified goals from the general courseware and the personalized needs and requirements as resulted from each learner’s profile [8].

Sessions have to be created in a modular fashion. Thus, a single session needs to be split into a hierarchy of elements, which are selected according to application-specific parameters and then combined into the final session protocol [7]. Upon connection to the e-learning service, some kind of a negotiation process should be invoked to reserve the required network parameters, while taking into account user’s desirable quality of service parameters as well as the current network conditions [5].

3.

E-learning management architecture

In order to design and implement our e-learning service management scheme, we will exploit experience gained from the area of network management, since this service is distributed and is provided via network routes.

(some) E-learning Application

Middleware (E-learning Manager)

Network Management

Component

Physical Network Infrastructure (Diverse Technologies) Learning Management Component Diverse Network-Communication Protocols

Expansion and complexity of telecommunication networks has induced growing challenges to efficiently manage the network elements according to the network providers and users objectives and expectations. As a result, a lot of research effort has been given in order to solve problems arising in this area and to establish standards that could be used across a broad spectrum of product types in a multivendor environment [10]. In response to the need for standards, Internet Engineering Task Force (IETF) has undertaken initiative to introduce a family of protocols that could be deployed in the area of networking, the well-known Simple Network Management Protocol (SNMP) [12].

The general architecture of the e-learning management scheme is based on the configurable middleware architecture for deploying e-learning services presented in [1], as well as on a client-server architecture, where the server is called agent, while the client is the managing process or manager. Each component has an agent, which maintains a local Management Information Base (MIB), and can communicate with the manager, residing in the service or session management component, through a management protocol such as the SNMP. The choice of the specific management protocol is apparent when considering that most e-learning services are implemented using Internet technologies. Therefore it is reasonable to choose protocols and standards that are adopted by the Internet community and are extensively deployed by diverse organizations, vendors, etc.

A MIB is a conceptual representation of all resources that provide the network manager with the ability to observe and control the current behavior of components and consequently of the whole e-learning service. Interaction between the manager and the management agents allows the retrieval and/or update of MIBs information in a way that enables the implementation of various management functions [9], from monitoring the service’s state to the flexible establishment and management of each user’s current session.

The deployment of MIBs provides the manager component with the ability to observe and control not only the static attributes of the managed components but also their current, dynamic, per session behavior. Agents monitor and control the e-learning service by retrieving and changing the values of their local MIB objects. The communication between the manager and an agent is based on a standardized network management protocol, like SNMP, which is usually deployed in TCP/IP networks, and their corresponding framework.

More specifically, the general e-learning service management architecture consists of the following main elements, as presented in Figure 2, as well:

Öa number of managed components, each containing an agent that maintains a local Management Information Base (MIB),

Öat least one manager component, and

Öa management protocol, such as the SNMP, which is used by the manager and the agents to exchange management information.

Figure 2. E-learning Management Scheme Architecture

An alternative implementation of an e-learning service, one, that favorites expansion; is a distributed service –as far as the learning server is concerned–. In this case, someone can consider a group of distributed learning servers like the one presented in Figure 2, where one server undertakes the role of a central manager of the whole service, who communicates and cooperates with all the other distributed managers, based on the SNMP. In next section, we further analyze all managed components.

4.

Managed components analysis

The prime component is the one that creates dynamically the profile of a session for a requested service (Session Management Component) by taking into account several parameters in order to be open and flexible. Each of these parameters is regarded as a different component of the middleware in a view to meet requirements for flexibility and implementation simplicity. This is the component that undertakes the role of manager in our management architecture. Each component might be comprised by several objects. For every object certain properties are User Agent Administrative Component Logging-Accounting Agent Policy Management Agent M I B Educational Process Management Component Agent M I B Educational Content Management Component Agent M I B Network Aware Component Agent M I B User Profile Management Component Agent M I B Session Management Component Agent M I B MANAGER : Interactions between Agent and local MIB : Interactions between Manager and Agents, based on SNMP

defined in order to describe required information (local MIB). One or more agents will be associated with each managed component.

User Profile Management Component

The same user might request the service with different session characteristics each time. Therefore, it is important that each session is built dynamically depending on the specific context. The corresponding agent of the component will be responsible for gathering all information regarding user’s profile.

User identification object properties [2]: ÖFirst name

ÖLast name ÖUser-Id ÖUsername ÖPassword

ÖRole (e.g. teacher, student, etc.)

Educational object (there is an entry for each course that

user has enrolled in –student– or is teaching –teacher) properties:

ÖCourse-Id

ÖPrevious knowledge level

Some elements concern not individual user but group of users. The values of the properties of these elements are dynamically updated every time a new session from a user or a group of users is initiated.

ÖNumber of simultaneous users of the group on-line ÖCooperative e-learning process for the group

(enabled or not)

Educational Process Management Component

Someone can identify three main types of e-learning services depending on time limitations and the allowed amount of interactions and cooperation.

ÖType of e-learning service

We can identify three values for this attribute: • Asynchronous self-paced e-learning • Asynchronous cooperative e-learning

• Synchronous cooperative e-learning or Virtual Classroom

Educational Content Management Component

Educational Content Management Component will be aware of the available types of educational content each time. It will be also aware of content’s specific characteristics. It is the component that will be responsible for choosing among different versions of the same content, depending on transfer and presentation requirements. Therefore, for each course the following properties have been identified:

ÖCourse name ÖCourse-Id ÖSubject area

ÖPrerequisite knowledge level (elementary, intermediate, advanced)

ÖCourse duration

ÖMinimum required bandwidth ÖSuggested bandwidth

ÖObjects comprising the course (several objects might comprise the educational content offered within a course). Each object will be associated with a set of properties:

• Object name

• Version index (in order to support different versions of the same content)

The following properties of an object are forming a

record that will have as many occurrences in the

MIB as the number of different versions of the object:

• Version-Id • File name

• Format (classification of content in specific type according to MIME)

• Location • Size

• Required transmission bandwidth

Network Aware Component

This is the component that will be aware of the whole underlying network infrastructure. It will also manage every session-related communication context. Network Component will also be aware of all available diverse physical network technologies supporting the connection and communication with users. It will also be aware of network protocols supported or required by different educational processes and educational content in order for them to be delivered to users. We can suppose that all this information does not change in a very dynamic way, but network component will be responsible for managing the whole session’s communication issues, so a network agent detecting the current conditions of network is required. This agent will be responsible for controlling all communication with user, throughout the whole session. The MIB of this component will comprise the following properties with relatively static values:

ÖSupported network technologies (e.g. PSTN or ISDN dialup connection, Fast Ethernet connection, Gigabit Ethernet connection, ATM connection, xSDL connection, cable connection, etc.)

ÖSupported network protocols (e.g. IP, TCP, UDP, RTP, etc.)

ÖCapacity of e-learning server’s connection

It will also comprise properties with dynamic values, calculated on real-time:

ÖNumber of active sessions at specific time ÖCurrent available bandwidth for next session

Administrative Component

This is the component that manages all administrative issues arising from providing an e-learning service. This component will also be responsible for managing the

“Service Contract” between user and service provider. This contract will define the e-learning service that user has agreed to receive.

Policy management object ÖPricing and charging policy

Pricing policy criteria: • Session duration • Type of service

• Type of educational content • Course

ÖAuthentication/authorization process (e.g. simple username/password process, certificates issuing process, etc.)

ÖRegistration procedures ÖTime constraints

Logging–Accounting Sub-Component

This Sub-Component will manage the logging and accounting procedure. Whenever a new e-learning session starts, an accounting process for the specific user should be triggered. Eventually, the whole session should be logged depending on certain parameters, not only for the shake of administration but also for user evaluation. The MIB of this component will comprise the following properties: ÖUser-Id ÖSession-Id ÖSession duration ÖType of service ÖCourse-Id

ÖObjects (Object-Id) that were accessed during session

Session Management Component

It is the main component of the middleware, the one that manage interactions of all others and the one that is responsible for the orchestration and management of a session. It is responsible for initiating or rejecting a new session with a user, while at the same time acts as some kind of broker between the user and e-learning service itself during the whole session. It undertakes the manager’s role in the e-learning service management process. This means that, Session Manager is responsible for gathering all required information from the agents of all managed components, in order to manage the e-learning service. It also manages in a dynamic way every

user session based on the values of the following

attributes: ÖSession-Id ÖUser-Id

ÖCurrent connection bandwidth

ÖCurrent network connection technology ÖUser-IP

5.

Conclusions

This proposed e-learning service management architecture is a first approach to the problem. We are currently working on the formal definition of MIBs based on the ASN.1. We also focus on standardisation issues concerning the implementation of such a management scheme. Future work will also explore formal ways of defining interactions, cooperation and communication among diverse components. Currently, we are in the process of designing and implementing a prototype of the proposed architecture. This prototype will also allow us to further define integration with existing e-learning systems or other e-learning architectures under development.

6.

References

[1] T. Apostolopoulos, A. Kefala, “A configurable middleware architecture for deploying e-learning services over diverse communication networks”, submitted in International Conference.

[2] T. Apostolopoulos, A. Kefala, “Developing distant learning services over broadband networks: concepts and alternatives”, IASTED International Conference Internet and Multimedia Systems Applications (IMS ’99), Conference Proceedings, 1999, pp.197-202.

[4] Y. Lee, Q. Chong, “Multi-agent systems support for community-based learning”, Interacting with Computers, Vol. 15, 2002, pp. 33-55.

[5] C. Bouras, V. Kapoulas, D. Miras, V. Ouzounis, P. Spirakis, A. Tatakis, “On-demand hypermedia/multimedia service using pre-orchestrated scenarios over the Internet”, Networking and Information Systems Journal, Vol.2, No.5-6, 1999, pp. 740-762. [6] M. Clarke, G. S. Blair, G. Coulson, N. Parlavantzas, “An efficient component model for the construction of adaptive middleware”, Middleware 2001, LNCS 2218, 2001, pp.160-178. [7] B. Stiller, C. Class, M. Waldvogel, G. Caronni, D. Bauer, “A flexible middleware for multimedia communication: design, implementation and experience”, IEEE Journal on Selected Areas in Communications, Vol.17, No.9, September 1999, pp. 1580-1598.

[8] C. Frasson, L. Martin, G. Gouarderes, E. Aimeur, “LANCA: A distance learning architecture based on networked cognitive agents”, ITS ’98, LNCS 1452, 1998, pp.594-603.

[9] T. Apostolopoulos, V. Daskalou, “Temporal Network Management Information Model and Services”, Journal of Networks and Systems Management, September 1998, Vol. 6, No. 3.

[10] T. Apostolopoulos, V. Daskalou, “Temporal Network Management Model: Concepts and Implementation Issues”, Computer Communications, 1997, Vol. 20.

[11] T. Apostolopoulos, V. Daskalou, “Network Management Services using a Temporal Information Model”, Proceedings of IM’97, 1997, San Diego.

[12] J. Case, M. Fedor, M. Schoffstall, J. Davin, “A Simple Network Management Protocol”, RFC 1157, DDN Network Information Center, SRI International, May 1990.