Robinson Udechukwu's Webpage


PhD Student at North Carolina State University

North Carolina State University Logo

Email: rnudechu@ncsu.edu
Phone Number: 704-900-3654

  • Passive Pricing Algorithm: Flat-rate pricing algorithm
  • Weighted Critical Resource Pricing Algorithm: Responsive pricing algorithm that adjusts pricing penalties placed on resources based on the network's usage and the computed Strongly Connected Components (SCC) graph using Tarjan's Algorithm. The relationship between the link's OPS reservation and the weighted critical threshold, decides whether or not a network link is included in the computation of SCC graph. Links connecting SCCs are priced with a penalty while links within a SCC are granted a discount.
  • Network Blocking Projection oriented Pricing Algorithm: Responsive pricing algorithm that adjusts pricing penalties placed on resources by measuring the impact each feasible path has on the network blocking of the system, by comparing the change to the system (Current residual network vs. residual network with the feasible path applied). The average network blocking projection of the presented list of path alternatives decides which path alternative is given a rebate/premium to their base cost.
  • Customers with a pricing sensitivity is referred as being price conscious and those without are referred to as being price unconscious. Price sensitivities are reflected in their choice policies as a budgetary constraints and lack of a price sensitivity is viewed as a budgetary constraint of infinity. Choice Policy resolves as the intersection to both the budgetary constraint and the customer's selection preference.
  • Cheapest Available Preference: Selects the path alternative with the cheapest price
  • Premium Available Preference: Selects the most expensive path alternative
  • Random Selection Preference: Selects a random path alternative
  • Delay Sensitive Preference: Selects the path alternative with the lowest delay
  • Maximum Delay-Utility Preference: Selects the path alternative with the highest Delay-Utility value above a given threshold.

Delay-Utility curves are based on delay curves mentioned in "A Utility-based QoS Model for Emerging Multimedia Applications".
In the simulation study, each source-destination paths' delay is normalized with respect to both their shortest path and longest path alternative, to resolve into an utility ranking.
Resource Provisioning Approaches
Path Provisioning Strategies
Pricing Algorithm
Old Tests Run consisting of individual run with fixed population distribution using a single traffic trace per topology
Simulation Type: Only Circuit resources
Provider Strategies: CRPS, PPS, NBPPS-BnA
Description: 1000 simulated calls with exponential distributed arrival times (16) and holding time (32) from any random source or destination. Magnitude of requested bandwidth uniformly distributed between 1 - 100. Diverse population in this old test run had fixed chunk of populations not distributed across the call requests, instead customer policies were made proportionately to the population distribution. Each topology has a population distribution of either:
  • 100% Cheapest Customers
  • 100% Premium Customers
  • 100% Random Customers
  • 50% Cheapest, 50% Premium Customers
  • 25% Cheapest, 37.5% Premium, 37.5% Random Selection Customers
Measurements: Measurements


Test 1: 30 tests for each 14NSFnet, 24USnet, 19EON
Simulation Type: Only Circuit resources
Provider Strategies: CRPS, PPS, NBPPS-BnA
Description: 5000 simulated microseconds of calls with exponential distributed arrival times (4.8) and holding time (15) from any random source or destination. Magnitude of requested bandwidth uniformly distributed between 1 - 100. Each topology has a population distribution of either:
  • 100% Cheapest Customers
  • 100% Premium Customers
  • 100% Random Customers
  • 50% Cheapest, 50% Premium Customers
  • 33% Cheapest, 33% Premium, 33% Random Selection Customers
Measurements: Measurements
Summary: Summary


Test 2: 30 tests for each 14NSFnet, 24USnet, 19EON
Simulation Type: Only Circuit resources
Provider Strategies: CRPS, PPS, NBPPS-BnA
Description: 5000 simulated microseconds of calls with exponential distributed arrival times (4.8) and holding time (15) from any random source or destination. Magnitude of requested bandwidth 25% chance of selecting 10, 20, 50, 100. Each topology has a population distribution of either:
  • 50% Cheapest, 50% Premium Customers
  • 33% Cheapest, 33% Premium, 33% Random Selection Customers
Measurements: Measurements
Summary: Summary


Test 3: 50 tests involving only 14NSFnet
Simulation Type: Only Circuit resources
Provider Strategies: CRPS, PPS, NBPPS-BnA
Description: 5000 simulated microseconds of calls with exponential distributed arrival times (4.8) and holding time (15) from any random source or destination. Magnitude of requested bandwidth 25% chance of selecting 10, 20, 50, 100. Each topology has a population distribution of either:
  • 50% Cheapest, 50% Premium Customers
  • 33% Cheapest, 33% Premium, 33% Random Selection Customers
Measurements: Measurements
Summary: Summary
Notes: Test the variability found in Test 2's 14 node NSFnet: Mix2 confidence interval


Test 4: 30 tests for each 14NSFnet, 24USnet, 19EON
Simulation Type: Only Circuit resources
Provider Strategies: CRPS, PPS, NBPPS-BnA
Description: 5000 simulated microseconds of calls with exponential distributed arrival times (4.8) and holding time (15) from any random source or destination. Magnitude of requested bandwidth 25% chance of selecting 10, 20, 50, 100. Each topology has a population distribution of either:
  • 75% Cheapest, 25% Premium Customers
  • 25% Cheapest, 75% Premium Customers
  • 25% Cheapest, 25% Premium, 50% Random Selection Customers
Measurements: Measurements
Summary: Summary


Test 5: 31 tests for each 14NSFnet
Simulation Type: Switching nodes imitate OPCInet resources
Provider Strategies: CRPS, PPS, NBPPS-BnA
Description: 360000.0 simulated nanoseconds of calls with exponential distributed arrival times (330) and holding time (1500) from any random source or destination with packet traffic following Poisson distribution with an inter-arrival mean of 33 nanoseconds. Magnitude of requested circuit resources have a 25% chance of selecting 1, 2, 5, and 10 wavelengths. Packet requests have a 33% chance of selecting 64, 1500, 9600 byte packets. Each topology has a population distribution of either:
  • 33% Cheapest, 33% Premium, 33% Random Selection Circuit Customers
Packet Route Selection Strategy: Random Selection without defection
Measurements: Measurements