Recently, as Internet users and Internet-related services has increased explosively, high bandwidth has been desired. Thus, Optical burst switching (OBS) was proposed as a new switching paradigm for optical network.
OBS network usually employs one-way...
Recently, as Internet users and Internet-related services has increased explosively, high bandwidth has been desired. Thus, Optical burst switching (OBS) was proposed as a new switching paradigm for optical network.
OBS network usually employs one-way reservation, which sends a burst control packet (BCP) with a specific offset time before transmitting each data burst frame (BDF). Due to such a property, multiple bursts contend with the same wavelength on the same output link simultaneously and lead to high burst losses. And the contention occurs more frequent as the offered load increases. So, for the better resource utilization of the network, it is required efficient control scheme to resolve contention.
Therefore, segmentation based congestion control scheme was introduced in OBS network. This scheme is able to regulate the congestion according to segmentation sizes. However, because OBS network employs one-way reservation, segmentation occurs in spite of a low offered load. Therefore, if we only use the information of segmentation sizes, the network state may be detected as congestion even though it is in normal case. So this paper proposes a congestion control scheme to improve the detection accuracy by analysing and studying the characteristics of segmentation information through the AIS(Artificial Immune System).
The proposed scheme is processed as the following.
Firstly, we define the data expression using extracted the segmentation feature through analyzing the segmentation information about normal/abnormal state. In order to reduce the complexity of congestion detection, we count the number of the 0s after executing AND operation among the continuously transmitted segmentation information. At the same time, to improve the detection accuracy, we count the number of the 0s among the continuously transmitted segmentation information.
Secondly, we define the matching rule which discriminates between normal and abnormal based on Euclidean distance.
Thirdly, we generate the detector set using Negative Selection algorithm considering the frequency of data pattern, especially. On the other hand, since the network status are dynamically changed, if congestion judgement that only depends on the detector set based on Negative Selection algorithm, false detection occurs frequently. In order to compensate these problems, we utilize fuzzy number theory which can infer the degree of threat. The degree of threat is calculated by monitoring the number of alarm signals occurrence time periodically. The performance of proposed scheme is evaluated through the OPNET simulation. Simulation results shows that our scheme efficiently detected congestion states.