Optical Network Management Perspective for Quality Assurance

1
Optical Network Management Perspective for
Quality Assurance
Dr. Ardian Greca Yamacraw Assistant
Professor Department of Computer Sciences Georgia
Southern University
naidrag_at_ieee.org
Phone [phone]
2
Introduction (The impact of network failures)
Conclusion A single cable cut can lead to a
dramatic amount of lost traffic
3
Increasing Traffic
4
Some failure rates
5
Transmission Networks
6
Core Transmission Networks
7
Network Protection Ring
8
Network Protection Mesh Network
9
WDM Transport Network Model
10
Proposed TMN based Network Survivability
Management System
11
Proposed System Components

• NSMS should be able
• to evaluate the response procedures
• replan the response procedure when the network
  topology changes,
• update restoration techniques,
• or if the required survivability requirement is
  not met yet
• analyze the performance
• failure the recovery can be static or dynamic.
• The static technique performs the recovery by
  using the pool capacity strategy.
• The dynamic recovery uses more intelligent
  techniques and leads to effective local
  reconfigurations.
• NSMS has a spare resource allocation control
  function to ensure efficient resource utilization

12
Proposed System Components

• Network survivability status monitor
• Surveillance
• detect the NE under failure
• which part and layer are currently controlling
  the restoration process
• how the restoration process deal with the
  failure
• Control
• control the number of connections that are
  rerouted and where are rerouted
• which connection should not be rerouted at all
• Test
• initializes a survivability test (i.e., simulate
  failures
• do not interrupt the network!
• Evaluation
• evaluates survivability process

13
Survivability Control System

• The survivability control system (SCS) is a very
  important function of NSMS, because there are
  several failure types and restoration techniques
  for a network layer.
• SCS will respond with the right restoration
  technique(s) to a failure, since the performance
  of one technique might not meet the overall
  survivability-cost requirements of the network.

14
Survivability Control System

• Restoration technique function,
• Recover the failure inside one layer without
  affecting other layers
• Can use different techniques
• Priority restoration function and
• Choose appropriate technique
• Cost effective technique
• Alteration control.
• Use parallel activation
• Use sequential activation

15
Protection in optical layer
16
Recovery on upper layers
17
Comparison between two activation mechanisms

• Parallel activation
• Advantages
• fast restoration, because there is no need to
  coordinate different techniques,
• Disadvantage
• could lead to inefficient resource utilization
• causes resource waste after restoration.
• NSMS should update spare resources in the
  network, but unfortunately this process is very
  complex and can lead to a disruption of new
  incoming calls.
• Sequential activation
• Advantages
• does not have the above mentioned problems
• can be easily controlled,
• Disadvantages
• might lead to a longer restoration time

18
Timing issues

• time to detect the failure at layer i is TDi.
  This time, TDi, for the lowest layer determines
  the time to detect the failure,
• elapsed time to generate an alarm indication
  message from layer i to the layer i1 is TAi,
• estimated restoration time for a given
  restoration technique in a given layer i is TRi,

19
Using Agents

• Run computation- and communication- intensive
  applications in real time
• End users interact with their applications when
  they are running
• Applications
• Telecommunications
• Interactively steered high performance
  computations
• Data mining
• Distributed interactive simulations
• Smart sensor and instruments
• Math calculations
• Etc

20
Agents vs. conventional object-based
representations

• Why?
• Re-use without Re-work
• Increased throughput (provides parallel
  processing)
• On-line reconfigured applications in response of
  user request or user behavior change
• End users do not need to be aware of the current
  task representation
• Information obtained can be personalized to the
  information seeker
• Increases trust since information gathered from
  different agents
• Know where to get the information
• Better interaction
• Etc

21
Agent Cooperation

• the agent management platform will consist
• on the functions of network task management,
• agent task scheduling,
• communication and agent management
• agent generation,
• dispatch,
• clone,
• registration and dismissal
• Agents will be composed into sequent queue or
  parallel sub-tasks.
• sub-tasks will be provided to agent scheduling
  function to perform the task schedule for
  multi-agents.
• Agents will be produced and assembled with the
  service code and dispatched to perform a certain
  job.
• free agents needed by the busy agent at the
  target node to help the busy agent to accomplish
  its task.

22
Agent Cooperation

• From the network management requirements point of
  view, we design two kinds of intelligent agents.
• One we call the common agent and
• the other the scheduling agent.
• Both agents will be equipped with knowledge
  database where they can store the information
  they got from the environment.
• Agents communicate with each other or with the
  management platform through communication
  function.
• Scheduling agents have a scheduling code to
  manage all agents in their groups.
• Scheduling agents can give some new service code
  to free agents and make them to perform a new
  task.

23
Conclusions

• Propose a network survivability management system
  as a new independent function structure, focusing
  on its architecture.
• NSMS compound functions are established and
  described
• NSMS will preplan and download to every NE the
  response procedure, by coordinating different
  restoration techniques in different layers.
• uses the global survivability strategy
• sequential activation mechanism to construct
  cost-effective survivable WDM networks.
• The management function is based on the
  multi-agent framework,
• and a coordination protocol is also proposed.
• Future work
• an integrated spare resources control that
  considers different restoration techniques,
• and how to determine the best restoration
  techniques for a given failure.
• strongly related with the network size, traffic
  pattern, and the network cost.
• to analyze problems for communication between
  NSMS and agents when the network is heavily
  loaded.