Performance analysis and prediction of physically mobile systems

advertisement

Performance analysis and prediction of physically mobile systems

• Point view:

• Computational devices including Mobile phones are expanding.

• Different infrastructure from traditional systems.

• Mobile systems require connectivity, dynamicity and resource availability.

• Logical mobility (software mobility) and physical mobility

• (hardware mobility).

• Authors present a methodology for modeling performance of physically mobile systems .

• Introduction

• Mobile phones interact with other phones (peer to peer mode e.g. BLUETOOTH) and with fixed network backbone e.g. Cellular Connection in dynamic way.

• Mobile phones offer all functionalities .

• Mobile phones resources are very limited (little memory,

• Limited energy)

• Software engineering for distributed systems is inadequate

• To deal with dynamically environment for mobile systems

• Authors methodology

• They suggest a methodology for modeling performance of physically mobile systems as following:

1. Modeling of the application

• 2. layered queuing network (LQN) generation and performance analysis :

• (a) Meta-LQN generation

• (b) LQN models generation

• (c) physical mobility ( PhM) pattern characterization

• 3. Results interpretation

• 1)modeling of application

• a) use-case diagram

• b) component diagram

• C) sequence diagram

• Physical mobility description:

• To describe the physical mobility (PhM) patterns we use

UML State Diagrams where each node represents a context and the arrows among states represent the probability that the user will be moving from the starting context to the destination one

• 2.Meta-LQN Generation :

• 1

. the operational profile

• 2. the scheduling policy of software components

• 3. the loop repetition factors and behavioral alternative probabilities

• 4. the host demand

• 3.LQN Models Generation

it identifies the hardware components in the deployment

• diagram and instantiates an LQN devices for each of them;

it adapts the meta-LQN model according to the software

• components reachable/visible in the location.

it adds LQN tasks to LQN device interconnections according to the Deploy association in the deployment diagram.

it adds additional LQN tasks to LQN device interconnections according to the resource name executing the external operation the additional task

• 4.PhM Patterns Characterization

• Analysis Scope and Results Interpretation

• Predictive performance analysis becomes of primary importance in the context of mobile applications, given the

• dynamicity of these systems and the often scarce resources involved. The performance indices of interests in mobile applications are mainly service response time and device utilization.

• The utilization of devices, in particular of the bandwidth

• in wireless networks, can be extremely useful for the

• identification of the hardware performance bottlenecks and

• to evaluate potential alternatives, including the strengthening of the hardware

• Technical review

• 1)good new methodology, but still reqiures improvement

• 2)all environments of mobile systems are not considered like (connection effort, handover, etc…)

• 3)verr short description of methodology

• 4)the mathematical formulas are not illustrated well, just within the figure.

Download