TY - CHAP
T1 - Performance modelling and traffic characterisation of optical networks
AU - Mouchos, Harry
AU - Tsokanos, Athanasios
AU - Kouvatsos, Demetres D.
N1 - Harry Mouchos, Athanasios Tsokanos, Demetres D. Kouvatsos, 'Performance modelling and Traffic characterisation of optical networks', in Demetres Kouvatsos Ed., Network Performance Engineering, (Heidelberg, Springer, 2011), ISBN: 9783642027413
PY - 2011
Y1 - 2011
N2 - A review is carried out on the traffic characteristics of an optical carrier's OC-192 link, based on the IP packet size distribution, traffic burstiness and self-similarity. The generalised exponential (GE) distribution is employed to model the interarrival times of bursty traffic flows of IP packets whilst self-similar traffic is generated for each wavelength of each source node in the optical network. In the context of networks with optical burst switching (OBS), the dynamic offset control (DOC) allocation protocol is presented, based on the offset values of adapting source-destination pairs, using preferred wavelengths specific to each destination node. Simulation evaluation results are devised and relative comparisons are carried out between the DOC and Just-Enough-Time (JET) protocols. Moreover parallel generators of optical bursts are implemented and simulated using the Graphics Processing Unit (GPU) and the Compute Unified Device Architecture (CUDA) and favourable comparisons are made against simulations run on general-purpose CPUs.
AB - A review is carried out on the traffic characteristics of an optical carrier's OC-192 link, based on the IP packet size distribution, traffic burstiness and self-similarity. The generalised exponential (GE) distribution is employed to model the interarrival times of bursty traffic flows of IP packets whilst self-similar traffic is generated for each wavelength of each source node in the optical network. In the context of networks with optical burst switching (OBS), the dynamic offset control (DOC) allocation protocol is presented, based on the offset values of adapting source-destination pairs, using preferred wavelengths specific to each destination node. Simulation evaluation results are devised and relative comparisons are carried out between the DOC and Just-Enough-Time (JET) protocols. Moreover parallel generators of optical bursts are implemented and simulated using the Graphics Processing Unit (GPU) and the Compute Unified Device Architecture (CUDA) and favourable comparisons are made against simulations run on general-purpose CPUs.
KW - bursty traffic
KW - Compute Unified Device Architecture (CUDA)
KW - Dense-wavelength Division Multiplexing (DWDM)
KW - Generalised Exponential (GE) Distribution
KW - Generalised Exponential Distribution (GE)
KW - Graphics Processing Unit (GPU)
KW - Just Enough Time (JET) protocol
KW - Long-Range Dependence (LRD)
KW - Optical Burst Switching (OBS)
KW - optical burst switching (OBS) protocol
KW - Optical Packet Switching (OPS)
KW - Parallel Processing
KW - self-similar traffic
KW - Self-Similarity
KW - Synchronous Optical Networking (SONET)
KW - Wavelength division multiplexing (WDM)
KW - Wavelength Division Multiplexing (WDM)
UR - http://www.scopus.com/inward/record.url?scp=79955801571&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-02742-0-37
DO - 10.1007/978-3-642-02742-0-37
M3 - Chapter
AN - SCOPUS:79955801571
SN - 9783642027413
VL - 5233
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 859
EP - 890
BT - Network Performance Engineering: A Handbook on Convergent Multi-Service Networks and Next Generation Internet
PB - Springer Nature
CY - Heidelberg
ER -