Efficient Routing and Wavelength Assignment in Wavelength-Routed Optical Networks Johannes Hamonangan Siregar Doctoral Program in Policy and Planning Sciences,

Efficient Routing and Wavelength Assignment in Wavelength-Routed Optical Networks

Johannes Hamonangan SiregarDoctoral Program in Policy and Planning Sciences, University of Tsukuba,

1-1-1 Tennoudai, Tsukuba-shi, Ibaraki 305-8573, JapanEmail: [email protected] Tel:+81-29-853-5587

Hideaki TakagiVice President, University of Tsukuba

1-1-1 Tennoudai, Tsukuba-shi, Ibaraki 305-8577, JapanEmail: [email protected] Tel:+81-29-853-2005

Yongbing ZhangInstitute of Policy and Planning Sciences, University of Tsukuba

1-1-1 Tennoudai, Tsukuba-shi, Ibaraki 305-8573, JapanEmail: [email protected] Tel:+81-29-853-5071

(2)APNOMS 2003

Introduction

• Wavelength division multiplexing (WDM) optical network offers a great potential for future high speed applications in large-scale networks because of its wide bandwidth and high-speed data transmission

• The optical communication path between a pair of a source and a destination is called a lightpath

• We consider the routing and wavelength assignment (RWA) for large-scale WDM optical networks where each transmission request is served by an all-optical lightpath without wavelength conversion

(3)APNOMS 2003

RWA Problem

To establish a lightpath, we need to determine– The path (route) from the source to destination– Assignment of a wavelength to the path

• Static lightpath establishment problem– The set of connection requests is known in advance– The objective is to minimize the number of wavelengths used

• Dynamic lightpath establishment problem – Connection requests arrive to the network dynamically– The objective is to minimize the connection blocking probability

We consider the static lightpath establishment problem

(4)APNOMS 2003

RWA Algorithms

Previous works: Longest first fixed path (LFFP) algorithm by Chlamtac et al. IEEE

Trans. Comm., 1992. They use only fixed shortest paths for all s-d pairs and assign a wavelength to the longest path first

Minimum number of hops (MNH) algorithm by Baroni and Bayvel, IEEE/OSA JLT, 1997. They use alternate shortest paths to decrease the heaviest load and assign a wavelength to the longest path first

Our algorithms: Longest first alternate path (LFAP) . We use alternate paths for s-d pairs

that cannot be established by shortest paths only and assign a wavelength to the longest path first

Heaviest path load deviation (HPLD). We determine the initial lightpaths using LFFP and then deviate the path load for some s-d pairs that pass through the heaviest link to minimize 　 the number of wavelengths

(5)APNOMS 2003

LFAP Algorithm

The RWA problem is formulated as a knapsack problem as follows:

(6)APNOMS 2003

HPLD Formulation

(7)APNOMS 2003

Flowchart of Our Algorithms

(8)APNOMS 2003

Results of Previous Algorithms

4

Number of wavelengths required = 6 Number of wavelengths required = 4

MNHLFFP

56

7

2

3

8

4

1

2

6

73

8

5

4

(s,d) lightpaths Wl

(4,5) 4-2-1-5

(4,7) 4-3-7

(5,8) 5-7-8 w1

(5,6) 5-6

(6,7) 6-7

(1,8) 1-2-4-8

(1,7) 1-3-7

(6,8) 6-7-8 w2

(3,4) 3-4

(5,7) 5-7

(2,6) 2-1-5-6 w3

(3,8) 3-4-8

(2,7) 2-1-3-7 w4

(1,4) 1-2-4 w5

(2,3) 2-1-3 w6

(s,d) lightpaths Wl

(4,5) 4-2-1-5

(1,7) 1-3-7 w1

(3,8) 3-4-8

(1,4) 1-2-4

(4,7) 4-3-7 w2

(6,8) 6-7-8

(5,7) 5-7

(5,8) 5-7-8

(2,7) 2-1-3-7 w3

(3,4) 3-4

(5,6) 5-6

(2,3) 2-4-3

(1,8) 1-3-7-8 w4

(2,6) 2-1-5-6

(6,7) 6-7

1

(9)APNOMS 2003

Result of Our Algorithms

Number of wavelengths required = 4 Number of wavelengths required = 4

3

15

6

73

2

4

8

15

6

3 7

2

4

8

(s,d) lightpaths wl

(4,5) 4-2-1-5

(4,7) 4-3-7

(5,8) 5-7-8 w1

(5,6) 5-6

(6,7) 6-7

(1,8) 1-2-4-8

(1,7) 1-3-7

(6,8) 6-7-8 w2

(3,4) 3-4

(5,7) 5-7

(2,6) 2-1-5-6 w3

(3,8) 3-4-8

(2,7) 2-1-3-7

(1,4) 1-5-7-8-4 w4

(2,3) 2-4-3

(s,d) lightpaths wl

(2,3) 2-4-8-7-3

(2,6) 2-1-5-6

(3,4) 3-4 w1

(5,7) 5-7

(6,7) 6-7

(1,4) 1-5-7-8-4

(2,7) 2-1-3-7 w2

(5,6) 5-6

(4,5) 4-2-1-5

(1,7) 1-3-7

(3,8) 3-4-8 w3

(5,8) 5-7-8

(1,8) 1-2-4-8

(6,8) 6-7-8 w4

(4,7) 4-3-7

LFAP HPLD

(10)APNOMS 2003

Kanto Network

(11)APNOMS 2003

Comparison of Algorithms

Number of wavelengths Computation time

(12)APNOMS 2003

Conclusion

• LFAP and HPLD yield a less number of wavelengths than LFFP and MNH by using not only alternate shortest paths

• LFAP and HPLD provide less computational complexity than MNH, because– LFAP assigns a wavelength to the longest path first

– HPLD deviates the load of the heaviest path to the lightest paths