Vous êtes sur la page 1sur 4

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com Volume 3, Issue 1, January February 2014 ISSN 2278-6856

New Area in Optical communication: Optical Burst Switching (OBS) Networks


Reza Poorzare1, Shahram Jamali2 and Asgarali Bouyer3
1

Young Researchers Club, Ardabil Branch, Islamic Azad University, Ardabil, Iran
2 Department of Computer Engineering University of Mohaghegh Ardabili, Ardebil, Iran

Azarbaijan Shahid Madani University Faculty of computer engineering and Information technology

Abstract: In this paper, we give an introduction to optical


burst switching (OBS) networks and compare the difference schemes given to improve the congestion control mechanism in OBS networks. Optical Switching Networks uses WDM (wavelength division multiplexing) to deal with the heavy traffic in the Internet. In this network, packets are assembled into the bursts and sent throughout the network. OBS networks are bufferless, so burst contention is one of the most important problems in the network. Because of this contention, TCP cannot distinguish that packet drop is due to Congestion or contention, as a results throughput of the TCP is decreased heavily.

corresponding control packet is sent to reserve the resource in the networks and after a delay time called offset time the main burst is sent throughout the network. OBS networks use one-way signaling reservation algorithm. It means when a control packet reserve the resources in the intermediate nodes, it does not send an acknowledgement message to the transmitter node. Besides the bufferless structure of the OBS, using the oneway signaling is another reason that leads to burst drop in the network [8].

2. RELATED WORKS
Keywords: Optical Burst Switching, TCP Vegas, Transport
Control Protocol (TCP).

1. INTRODUCTION
In recent years, a large number of studies have been conducted to modify TCP to the OBS networks. The majority internet traffic is carried by TCP, so it is important to modify TCP to optical networks. TCP was designed for wired network which are using buffers, but OBS networks do not have any buffers, so when two nodes compete to reserve one wavelength, it causes burst contention. When a burst contention happens, TCP assumes that network is congested and decreases the flow rate wrongly and it leads to throughput reduction. The reason for this reduction is: sometime the traffic is light in the network but we have some contentions and decreasing the flow rate in these circumstances leads to the throughput reduction [1-7]. Optical Burst switching (OBS) is a switching technique that is used in the OBS networks to deal with the heavy traffic in the networks. The main reason of using this technique is to achieve to high performance in the networks. By using burst switching in the optical networks, bandwidth wasting is reduced. In OBS networks, packets are assembled into the bursts and sent throughout the network. The OBS networks use time/size based algorithm to make a burst. In time based algorithm after reaching to a threshold time making burst stops, but in size based algorithm when the size of the burst reaches to a threshold making bursts stops. OBS networks usually use both time/size based algorithm. Before sending a burst in the network, its Volume 3, Issue 1 January February 2014

False congestion is one of the most important problems in OBS networks. So a burst loss due to contention in OBS networks may cause throughput reduction. Packet drop in the OBS networks may not show congestion and it can show contention in the network, so a lot of researches have been done to differentiate burst drop by contention and burst drop by contention. Several approaches have been proposed to take advantage of optical communications in particular optical switching. One of these proposed schemes is optical circuit switching based on wavelength routing. In this scheme, lightpath needs to be established using a dedicated wavelength on each link from source to destination. Once the connection is set up, data remains in the optical domain throughout the lightpath. One of the other approaches is optical packet switching. In thing technique, while the packet header is being processed either all-optically or electronically after an Optical/Electronic (O/E) conversion at each intermediate node, the data payload must wait in the fiber delay lines and be forwarded later to the next node [9][10]. In order to provide optical switching for next generation, a new switching paradigm called optical burst switching (OBS) was proposed [11]-[12]-[13]. OBS networks exploit WDM techniques to deal with the heavy traffic of the internet. In this technique, each fiber channels are divided to several independent channels. Explicit signaling from the OBS layer to the TCP layer is proposed in [14] to cope with the TCP false congestion detection problem. Generating an explicit signal for every random contention in the OBS layer is the shortage of Page 270

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)


Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com Volume 3, Issue 1, January February 2014 ISSN 2278-6856
this method and it will significantly increase the network operation overhead and reduce the protocol practicality by breaking the TCP end-to-end semantics. Other approaches to solve the problem include burst retransmission and deflection schemes are proposed at the OBS layer [15]-[17]. The Proposed schemes can hide some burst loss events from the upper TCP layer, so it reduces the chance of TCP false congestion detection. With burst retransmission or deflection, bursts subject to contention can be retransmitted at an OBS edge or can be deflected to an alternate route in the OBS network, respectively. A novel scheme, threshold-based TCP Vegas, to achieve more precise network congestion identification in OBS networks under burst retransmission, is proposed in [18]. The proposed scheme adjusts the Congestion Window (cwnd) size based on the round trip times (RTTs) of packets received at TCP senders. If the number of RTTs that are longer than the minimum RTTs exceeds a threshold, the scheme detects network congestion, so is adjusts the cwnd size similar to conventional TCP Vegas; otherwise, the scheme adjusts the cwnd size based on the minimum RTTs. This Idea shows that a significant improvement is observed in terms of TCP throughput for the threshold-based TCP implementation over OBS networks with burst retransmission. One of the other schemes is a quantitative evaluation of reordering in OBS networks and its impact on TCP performance. Contention avoidance by balancing the load in the network contains two stages, route calculation and route selection. Route calculation can be implemented in a static or a dynamic manner. In static route calculation one or more routes are pre-calculated based on a static metric such as physical-distance or number of hops. On the other hand, dynamic route calculation routes are computed periodically based on certain dynamic traffic information such as link congestion or number of contentions. In the route selection stage, one of the routes which are calculated (either statically or dynamically) is selected in a static or a dynamic manner for the burst transmission. In static route selection, a fixed fraction of traffic is sent on all the alternate paths. In dynamic route selection, routes are selected based on certain dynamic feedback information. The dynamic feedback information is the congestion on the link and the congestion information is communicated back to the source periodically for every interval using either interlinkstate broadcasts or probe packets [19]. In OBS networks, bursts (assembled from packets arriving at the edge nodes) are sent to the destinations nodes through the core network. In intermediate nodes contention may happen and one of the burst will drop. A technique called burst segmentation is used at the core nodes for improving congestion control in the OBS networks [20]-[21]. Either the head of the contending Volume 3, Issue 1 January February 2014 burst, is the dropped segment here. In this case, number of received packet will be increased. Connection can be used to overcome contention losses and improve packet loss probability for improving recovery mechanisms for their burst transmission. A 1 + 1 protection mechanism to handle network component failures for OBS networks was proposed in [22]. In this scheme contentions are using dedicated channels for each connection. No other connections are allowed to share these dedicated bandwidth channels, statistical multiplexing is not possible. So, no bursts are lost and bandwidth could be underutilized. A new scheme called Forward Error Correction (FEC) were given in [23] in the context of providing recovery support for hard real-time communication. In this paper, FEC codes were used at the packet level, where redundant packets are transmitted along different routes as a form of protection. Another scheme was proposed in [24] called Forward error correction (FEC) for loss recovery. In this paper FEC is the base of the scheme. In the networks channels are unreliable and a lot of errors occur in them, so FEC can be used to solve this problem. In this scheme, if an error happens, the transmitting node is notified by the destination node to take necessary actions. Interconnection of TCP/IP networks in optical networks was considered in [25], given the worldwide spreading of TCP/IP based applications, with reference to the network scenario shown in Figure 1 The goal of this scheme is providing a framework for fast evaluation of the influence of OBS node and network design options on end-to-end performance. The problem of TCP flow synchronization in OBS networks was given in [26] and it was extended in [27]. In this paper, TCP flows in IP networks were explained then its impact on bandwidth consumption was discussed. Next, how an OBS network suffers from TCP synchronization was discussed.

Figure 1 The reference network layout. One of the other approaches that have been proposed to overcome the random burst contention in the network is called Forward Segment Redundancy (FSR). If we want to use burst segmentation in the network, we have to divide burst data into the number of segment, each of Page 271

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)


Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com Volume 3, Issue 1, January February 2014 ISSN 2278-6856
them which has a payload and a header. The segments may have one or more data packets. When a contention happens in the network, one or more than one segment are dropped in the network instead of dropping whole burst. Figure 2 shows an example of burst contention by using segmentation. This is the base idea that has been used in FSR. 4. Deflection routing may cause a loop to a burst, so a burst may never arrive to the destination. 5. In high traffic, deflection routing does not work properly [29].

3. CONCLUSION
In this tutorial we have first given an introduction to optical burst switching (OBS) networks. We had a comparison between this new paradigm and other switching techniques, and it showed OBS is a viable technique for the next generation optical networks. We have provided a brief historical review of the recent works on OBS networks. This tutorial has also attempted to provide a comprehensive coverage of research issues related to OBS. Among this issue, there are a lot of scheme that attempted to solve the false congestion control mechanism.

Figure 2 Burst drop by using segmentation FSR is an extension of segmentation scheme. The packets are stored in an electronic buffer as burst segments before each data burst is created. In this electronic buffer a copy of the burst is stored. Next, the ingrees node is responsible to make redundant data to each burst. The percentage of the burst is used to determine the size of the redundant. When a burst contention happens in the network, there are some rules to drop the segments. If a burst has redundant data it is considered as high priority, while a burst with no redundant data is considered high priority. The core nodes use the following policies: Combined head and tail drop (HTD): In this case, a part of the tail of the original burst and a part of the head of the contending burst are dropped. Head drop (HD): The head of the contending burst is dropped so that there is no longer contention. Tail drop (TD): The tail of the original burst is dropped so that there is no longer contention. Drop contending burst (DC): The entire contending burst is dropped and the original burst is scheduled [28]. Deflection routing was another approach that deals with the false congestion detection in the network. This scheme involves selecting a different output fiber than the intended one, if the burst cannot be switched through the original output fiber. This method has its own advantages and disadvantages. 1. The offset time needs to be recalculated, because the output has been changed and the deflected burst takes a different path. 2. Offset recalculation will need core nodes to be equipped with FDLs, so they can delay the deflected bursts if necessary. 3. In this method, bursts may arrive out of the destination order. So, destination nodes may have to store large amount of data. Volume 3, Issue 1 January February 2014

References
[1] A.Jain, S. Floyd, M.Allman, P.Sarolan Quick-start for TCP and IP, ICSI, 2006. [2] C. Jin, D. Wei, S. Low, FAST TCP: motivation, architecture, algorithms, performance, INFOCOM 2004. Twenty-third AnnualJoint Conference of the IEEE Computer and Communications Societies (Volume:4 ) , March 2004. [3] S. Hegde, et al., FAST TCP in high-speed networks: an experimental study, in: Proceedings, GridNets, Engineering & Applied Science, Caltech, the First International Workshop on Networks for Grid Applications, 2004. [4] L. Xu, K. Harfoush, I. Rhee, Binary increase congestion control (BIC) for fast long-distance networks, in: Proceedings, INFOCOM 2004. Twentythird AnnualJoint Conference of the IEEE Computer and Communications Societies (Volume: 4), March 2004. [5] W. Stevens, TCP slow start, congestion avoidance, fast retransmit,and fast recovery algorithms, RFC, 1997. [6] M. Mathis, J. Mahdavi, S. Floyd, A. Romanow, TCP selective acknowledgement options, RFC, 1996. [7] L. Brakmo, L. Peterson, TCP Vegas: end-to-end congestion avoidanceon a global internet, IEEE Journal on Selected Areas in Communication, 1995. [8] B. Shihada, P. Ho, Q. Zhang. SAIMD: A Novel False Congestion Detection Scheme in TCP over OBS Networks, a Lightwave Technology, Journal of (Volume:27 , Issue: 4 ), pp. 386-395, 2009 [9] D. J. Blumenthal, P. R. Prucnal, and J. R. Sauer, Photonic packet switches: architectures and experimental implementations, Proceedings of the IEEE, vol. 82, pp. 16501667, November 1994. [10] G.-K. Chang, G. Ellinas, B. Meagher, W. Xin, S.J. Yoo, M.Z. Iqbal, W. Way, J. Young, H. Dai, Y.J. Page 272

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)


Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com Volume 3, Issue 1, January February 2014 ISSN 2278-6856
Chen, C.D. Lee, X. Yang, A. Chowdhury, and S. Chen, Low Latency Packet Forwarding in IP over WDM Networks Using Optical Label Switching Techniques, in IEEE LEOS 1999 Annual Meeting, 1999, pp. 1718. [11] M. Yoo and C. Qiao, Just-enough-time (JET): A high speed protocol for bursty traffic in optical networks, in Proceeding of IEEE/LEOS Conf. on Technologies For a Global Information Infrastructure, August 1997, pp.2627. [12] C. Qiao and M. Yoo, Optical burst switching (OBS)a new paradigm for an optical Internet, Journal of High Speed Networks, vol. 8, no. 1, pp. 6984, 1999. [13] J. Turner, Terabit burst switching, Journal of High Speed Networks, vol. 8, no. 1, pp. 316, 1999. [14] X. Yu, C. Qiao, Y. Liu, TCP implementations and false time out detection in OBS networks, in: Proceedings, IEEE Infocomm, Hong Kong, China, March 2004. [15] Q. Zhang, V. Vokkarane, Y. Wang, J.P. Jue, Analysis of TCP over optical burst-switched networks with burst retransmission, in: Proceedings, IEEE GLOBECOM, St. Louis, MO, November 2005. [16] Q. Zhang, V. Vokkarane, Y. Wang, J.P. Jue, Evaluation of burst retransmission in optical burstswitched networks, in: Proceedings, 2nd International Conference on Broadband Networks, BROADNETS, Boston, MA, October 2005. [17] C. Hsu, T. Liu, N. Huang, Performance analysis of deflection routing in optical burst-switched networks, in: Proceedings, IEEE Infocomm, New York, NY, June 2002. [18] B. Shihada, Q. Zhang, P. Ho, Jason P. Jue, A novel implementation of TCP Vegas for Optical Burst Switched networks, Optical Switching and Networking, 2010. [19] B. Komatireddy, D. Chandran, V. Vokkarane, TCPaware Load-Balanced Routing in Optical BurstSwitched (OBS) Networks, Optical Fiber Communication and the National Fiber Optic Engineers Conference, 2007. OFC/NFOEC 2007. Conference on , 2007, pp. 1-3. [20] V. Vokkarane, J. Jue, Prioritized routing and burst segmentation for QoS in optical burst-switched networks, in: Proceedings of Optical Fiber Communication Conference (OFC), pp. 221222, 2002. [21] V. Vokkarane, J. Jue, Prioritized burst segmentation and composite burst assembly techniques for QoS support in optical burst-switched networks, IEEE Journal on Selected Areas in Communications 21 (7) (2003) 11981209. [22] D. Griffith, S. Lee, A 1 + 1 protection architecture for optical burst switched networks, IEEE Journal on Volume 3, Issue 1 January February 2014 Selected Areas in Communications 21 (9) (2003) 13841398. [23] S. Raghavan, G. Manimaran, C. Siva Ram Murthy, A note on dependable real-time communication in multihop networks, Computer Communications 25 (17) (2002) 16731683. [24] M. Casoni, C. Raffaelli, Analytical framework for end-to-end design of optical burst-switched networks, Analytical framework for end-to-end design of optical burst-switched networks, pp. 1039-1046, 2012. [25] M. Casoni, E. Luppi, M.L. Merani, Impact of assembly algorithms on end-to-end performance in optical burst switched networks with different QoS classes, in: Proceedings of IEEE/SPIE Third Workshop on Optical Burst Swithching, San Jose, CA, October 25, 2004. [26] O. Gonzlez de Dios et al., On transmission control protocol synchronization in optical burst switching, Photonic Network Communications, vol. 18, pp. 323333, 2009. [27] O. de Dios, I. de Miguel, R. Durn, J. Aguado, N . Merayo, P. FernndezImpact of TCP Synchronization on Capacity Dimensioning of Optical Burst Switched (OBS) Links, Networks and Optical Communications (NOC), 2012 17th European Conference on, pp. 16, 2012. [28] J. Sullivan, N. Charbonneau,V. Vokkarane, Performance Evaluation of TCP over Optical Burst Switched (OBS) Networks using Coordinated Burst Cloning and Forward Segment Redundancy, Global Telecommunications Conference (GLOBECOM 2010), 2010 IEEE , pp. 1-6, 2010. [29] X. Mountrouidou, V. Puttasubbappa, H. Perros, A Zero Burst Loss Architecture for star OBS NetworksNetwork control and engineering for QoS , security and mobility , pp. 1-8, 206.

Page 273

Vous aimerez peut-être aussi