Proceedings of 2nd National Conference on Challenges & Opportunities in Information Technology (COIT-2008)

RIMT-IET, Mandi Gobindgarh. March 29, 2008

Integration of Mobile Computing with Grid Computing: A Middleware

Architecture

Anil Kumar*, Salim Raza Qureshi **

Model Institute of Engg. & Techn, Kot Bhalwal, Jammu
*[email protected], **[email protected]

Abstract— In this paper we present a proposal of a This middleware integrates the tools of grid
middleware to integrate mobile computing with grid computing – e.g., data communication, directory and
computing. We describe the support to be provided by service distribution, security, resource discovery,
the combination of the two technologies, propose a and resource allocation – providing the developer a
middleware to support the development and management of
mobile grids, and discuss how this scenario contributes to
set of reusable and homogeneous resources.
the development of a new age of mobile service This paper is organized as follows: Section II provides
applications. the background of the problem. Section III proposes
system architecture and shows how it solves the problem.
I. INTRODUCTION The paper concludes with section IV.

The development of mobile services involves two II. ANALYSIS

problem domains: (i) the problem of the computational
infrastructure for the implementation of services, which Based on the above example, on the characteristics of the
regards the hardware and software structure that necessary computational infrastructure, and on
facilitates the connection of mobile devices (devices, data the implementation of the application for this
transport system, and others) and service base intrinsic to environment, we present the requirements for this problem
this scenario (location identification service, fault tolerance scenario:
services, and others) [5] and; (ii) the problem of service
implementation itself, in which the developer challenge is (a) The data communication methods, necessary for
to integrate the available computational infrastructure and data transmission from the sensors and integration of the
create a solution to the problem of the scenario in actuators and personal assistants; (b) the security
question. Furthermore, as described in [7], mobile methods for data transmission and access control to
computing imposes a degree of complexity inherent to service resources; (c) the method of resource registration
the environment, such as dynamic environments, that allows, for instance, the sensors or personal assistants
mobility, computational resource limitations, latency entering and exiting control in the system; (d) the device
and instabilities in data transfer, energy supply management methods that allows to manage sensors,
limitations, and input/output interface limitations. actuators, and personal assistants in a centralized
Several studies are being conducted and way, despite the distribution, environment
approaches are being proposed to circumvent these dynamism, and device heterogeneity, (e) the interface
problems. Amongst these, we can cite advances in the method of mobile devices that allows the communication
areas of (a) data communication, such as faster and more with the users or the connected devices.; and (f) the
reliable networks [8]; (b) distributed computing, through methods necessary for work coordination between
processing distribution among client and server [2]; and mobile users that allows collaboration between
(c) software engineering, with the creation of applications personal assistants.
that adapt their behavior according to the resource
limitations [4]. However, to date the proposed The support for these requirements is beyond the
methodologies don’t offer a complete solution, and in this scope of programming languages (e.g., Java 2 Micro
way they lead to ad hoc solutions, specific to the problem Edition [1]) and requires a software solution that
scenario being dealt. The lack of re-usability and achieves device integration, development homogenization,
homogenization causes incidental costs in the development of software reuse, coordination, resource discovery,
these applications. resolution of addressing problems, security, and other
characteristics.
We believe in the integration of grid computing and
mobile computing technology through a middleware for We believe there exists a relation between the
the development of mobile service applications. solutions provided by Grid Computing [9] and the
development of systems for the scenarios of mobile

104
Proceedings of 2nd National Conference on Challenges & Opportunities in Information Technology (COIT-2008)
RIMT-IET, Mandi Gobindgarh. March 29, 2008

computing, as described before. For instance, mobile Gridbus [10], developed by the Grid
computing requires solutions for service discovery in a Computing and Distributed Systems (GRIDS) Laboratory
dynamic environment, establishment of data from University of Melbourne, is a open source grid
communication channels between devices and service software package for the development of grids for
providers, development tools that allow the integration eScience and eBusiness. It uses several other
of devices and distributed resources, and security, middleware, such as Globus, Unicore [12], Alchemi
such as confidentiality of data and authentication. We [15], and others. Its support for mobile services
suggest that grid computing technology provide the tools characteristics is found in collaboration, resource
necessary for the implementation of these systems. allocation, and dynamic environments, provided by lower
level middleware or core middleware, such as the
Initially, we will analyze grid computing software e ones described before.
discuss the solutions provide by them when applied in the
development of mobile applications. This analysis Legion [11] is a middleware developed by a
will help in the identification of the shortcomings of the project at University of Virginia, and is defined as a
existent solutions and position our contribution. meta-system based on objects (resources) with billions
of hosts and trillions of objects linked together by high-
III. RELATED WORK speed networks, workstations, and supercomputers in a
system that can aggregate different architectures,
Grid computing technology offers support operating systems, and physical locations. The support
for the implementation of mobile services for provided by Legion for mobile service characteristics is
decision support systems. There are several seen in collaboration, in which the binding system makes
software packages for the implementation of grid possible the collaboration through tuples such as that
computing services. In this sub-section we will present supply object addressing (LOID) and
the ones most known by the community and discuss their management (LOA). There also exists resource allocation,
support for the creation of mobile services. provided by LOA (Legion Object Addresses) that
incorporates a physical address as the IP and can
As described in [6], the development of decision distribute these resources through multicast.
support mobile services needs the support of:
collaboration, user interface, context-awareness, and a UNICORE (UNIform Interface to Computer
inference process that allows the development of Resources) [12] is a middleware that integrates grid
systems beyond the purely reactive. In an environment of computing resources through a graphical interface
limited resources, applications must be aware of the developed in Java. Its support for mobile service
computational resources limitations, unreliable and characteristics can be found in collaboration provided
intermittent communication networks, energy supply through the UNICORE servers after the client/user
limitations, mobility, dynamic environments, and reduced authentication. The collaboration is achieved by the
user interfaces. These limitations are inherent to the servers sending jobs to peer Unicore gateways, which
environment and will be soothed, and not execute the jobs and respond. The support of resource
superseded, by future technological developments [7]. allocation is achieved by the Abstract Job Object (AJO),
Based on the requirements described before, we which is an object that sends and receives jobs
evaluated the existent grid middleware in terms of Table I summarizes the support provided by
collaboration support; context-awareness support; the middleware packages for the development of
resource allocation support; dynamic environments mobile computing services. From the related work
support; and execution on mobile devices support. analysis we conclude that the existent packages do
not satisfy the requirements for the creation of mobile
Globus [3] is a open source software package developed
services, such as the support of collaboration,
by the Globus Alliance which offers a application
context awareness, resource allocation, dynamic
development and grid computing systems toolkit. Its
environments, and execution on mobile devices.
support for mobile service characteristics is found in
collaboration through resource layer protocols which can Therefore, we identify the need for the creation of a
obtain information and also control jobs, Middleware for Mobile Devices that provides the
promoting collaboration and resource distribution. necessary functionality to support the development of
Also there is resource allocation provided through mobile service applications.
the resource manager (Globus Resource Allocation
Manager – GRAM) which supplies a job sending and
monitoring interface.

105
Proceedings of 2nd National Conference on Challenges & Opportunities in Information Technology (COIT-2008)
RIMT-IET, Mandi Gobindgarh. March 29, 2008

TABLE I. GRID MIDDLEWARE COMPARISON

Fig. 1 depicts the architecture. The (A) middleware
LEGI UNIC
GLOBUS GRIDBUS ON ORE serves as a bridge that links (B) applications running
Collaboratio on mobile devices and the (C, D) grid computing
n Yes Yes
support Yes Yes No No
services running on central grid management offices. In
Context Yes Yes these central offices, the (C) Directory Module stores
(LO (AJO service data in (S1) directory databases. The (D)
awareness No No A) )
support No No
Broker Module manages the resources, schedules jobs
Resource Yes No and monitors their execution, storing monitoring data and the
allocation Yes No
job results in (S2) task databases.
(GRA LEGI
M) ON
UNIC To satisfy the requirements listed in the previous
support Yes ORE
section, let’s consider the architecture, it’s
Dynamic No Yes
No No Yes No components, and the following conclusions:
(LO i) To satisfy the requirements of (a) data communication
environment A) Yes
(AJO
methods, the HTTP protocol is used,
support No ) guaranteeing compatibility between devices.
Mobile ii)To satisfy the requirements of (b) security in
device No No
data transmission and access control, authentication
No Yes
LEGI
methods and cryptography are used in node registrations
execution ON No iii) To satisfy (c) resource registration, the
support
Directory Module uses the LDAP [14] protocol and
maintains the resource registrations.
IV. PROPOSAL iv)To satisfy (d, f) device management and
coordination, the Broker Module manages the devices
In the previous section, we presented our motivation, using its management methods, which will be described in
the analysis of the requirements needed for the sub-section B.
implementation of applications that support mobile v)To safisfy the (e) interface with the devices,
services, and the necessary characteristics of a mobile XML technology [13] is used for data
grid. Consequently, the question that now arises is: representation and visualization at the devices.
A. System Architecture
B. Application Case Study
In this sub-section we present the elements and
interactions of our proposed System Architecture for the In this sub-section we elaborate a case study
integration of mobile devices in computational grids. of an application using the proposed architecture.
Consider the following scenario:
A doctor consults a patient data, whom is being
GRID-MOBILE monitored by a sensor. The sensor periodically sends the
ARCHITECTURE patient’s vital data for storage, which is made available
Directory Module to doctors.
(C) For this to happen, the following steps must be attended:
S (nodes, services Broker Module
1 management) (D) 2
Tasks i) Doctor’s PDA registration on the grid;
Sto
rage and ii) Previous sensor registration;
Tasks
No iii) Data request by the doctor;
Result
des+
s iv) Patient search and consulting of the attached sensors
Ser
Datab
vice
ase v) Data sent to doctor’s PDA.
HTTP-XML Server module
The following methods and architecture components
are involved in the described scenario:
ƒ In (i) and (ii) the method for authenticating the PDA
on the grid is registerNode(PDA_type) and the
Directory Module executes it.
ƒ In (iii) and (iv) the method for data request is
submitTask(searchPatient) which answers the

106
Proceedings of 2nd National Conference on Challenges & Opportunities in Information Technology (COIT-2008)
RIMT-IET, Mandi Gobindgarh. March 29, 2008

patientID with his characteristics and sensor - deleteNode(nodeName);

capabilities.
ƒ In (v) the method fetchTaskResults(taskid) sends Submit and Receive Tasks Methods
the data and takes into consideration the - submitTask(nodeName, taskAction, taskData);
limitations of the PDA which requested it. - checkTaskStatus (taskId);
- fetchTaskResults(taskid);
In the described scenario, we verify that the
resource homogenization is an advantage for the better Management Methods
management of the environment. In this context, being - getNodes();
each resource a grid node, the middleware has through
- getNodeStatus(nodeName);
its methods the ability to manage information and physical
resources. - setNode (Node, param);

The middleware is depicted in Fig. 2. Some of its

API methods that enable its interaction with the REFERENCES
environment are listed in the appendix.
[1] Java 2 Micro Edition (J2ME)
CONCLUSION web-site, sun corporation,
http://java.sun.com/j2me.
This work proposes an architecture in which its
main component, the middleware, adds value to the [2] S. Das and S. K. Das, editors. 5th International
problem of mobility in grids and provides Workshop on Distributed Computing, volume 2918
homogeneous grid resource management. As described in of Lecture Notes in Computer Science, Kolkata,
Section III, the proposal makes up to the India, December 2003.
requirements of mobility, management, data [3] I. Foster. Globus toolkit version 4:
communication, security, and interface, which we believe Software for service-oriented systems. IFIP In
are the basic requirements of any environment that ternational Conference on Network and Parallel
provides mobility, context-awareness, and computing,
anytime/anywhere information. As future work, the [4] E. Guigere. Java 2 Micro edition: The
middleware can be developed ultimate guide
[5] on programming handheld and embedded
devices. John
[6] Springer-Verlag LNCS 3779:pp 2-13, 2005.
[7] E. Guigere. Java 2 Micro edition: The
ultimate guide on programming handheld and
embedded devices. John Wiley and Sons, Inc., USA,
2001.
[8] K. Henricksen, J. Indulska, and A.
Rakotonirainy. Infrastructure for pervasive
computing: Challenges. In K. Bauknecht, W. Brauer,
and T. A. MÄuck, editors, Informatik 2001: Wirtschaft
und Wissenschaft in der Network Economy
Visionen und Wirklichkeit, volume 1 of
Tagungsband der GI/OCG-Jahrestagung, pages 214-
222, Universitat Wien, September 2001.
Figure 2. Middleware API and interactions

into a framework in which other modules such as p2p,

security etc can be added.

APPENDIX
* Middleware API: Methods

Generals Methods
- registerNode(nodeName, nodeType);

107