Subscribe Now Subscribe Today
Research Article
 

On the Resource Allocation in Optical Burst Switching Networks



A. Abid, F.M. Abbou and H.T. Ewe
 
Facebook Twitter Digg Reddit Linkedin StumbleUpon E-mail
ABSTRACT

Optical Bursts Switching (OBS) is a flexible yet efficient photonic switching paradigm. However, due to their one-way resource reservation mechanism, OBS networks experience high bursts (thus packets) loss rate. In OBS networks, the contention is resolved either by dropping one of the contending bursts, or more efficiently by dropping from one of the contending bursts only the parts that overlap with the other burst. In both situations, only one data source will suffer the data loss in favor to the other. In this study a simple yet effective network resource allocation algorithm is presented. With this algorithm, the dropped segments are selected evenly from both contending data bursts. Furthermore, the truncated data bursts (forwarded or deflected) are guaranteed to be larger than the minimum burst-length allowed by the network specification. Additionally, the algorithm is demonstrated to be practical if used in a mechanism to provide differentiated services to IP packets over an OBS network. A simulation is used to evaluate the performance of the algorithm.

Services
Related Articles in ASCI
Similar Articles in this Journal
Search in Google Scholar
View Citation
Report Citation

 
  How to cite this article:

A. Abid, F.M. Abbou and H.T. Ewe, 2007. On the Resource Allocation in Optical Burst Switching Networks. Journal of Applied Sciences, 7: 3057-3062.

DOI: 10.3923/jas.2007.3057.3062

URL: https://scialert.net/abstract/?doi=jas.2007.3057.3062

INTRODUCTION

All-optical networking concept is the key strategy to sustain the ever-increasing demands for higher throughput and faster switching of current and future data communication networks. Particularly, with the implementation of Wavelength Division Multiplexing (WDM) technology in the network backbone, Optical Burst Switching (OBS) (Qiao and Yoo, 1999; Battestilli, 2002) appears to be an appropriate data transfer technique for all-optical networks. OBS is an approach where the benefits of both packet-switching networks (Blumenthal et al., 1994; Cruz and Tsai, 1996) and circuit-switching networks (Chlamtac et al., 1992; Mei and Qiao, 1997; Yuan et al., 1997) are combined. OBS takes into consideration the limitation of the existing all-optical technology (i.e., limited processing power, the lack of efficient buffering techniques and the limited number of wavelengths per physical fiber coupled with the high cost of wavelength conversion) (Vincent et al., 1998). In OBS networks the ingress nodes generate control packets that are sent out-of-band into the network an offset time ahead of macro-packets. The macro-packets named Data Bursts (DB) are made up of various upper layers’ packets (e.g., IP packets, ATM cells, SONET and Frame Relay frames). The control packets configure the fabric switch of the core nodes and reserve the necessary network resources to accommodate the upcoming data bursts, which will be switched and transmitted entirely within the optical domain.

For various reasons the control packet may fail to reserve the full/part of the resources needed to establish an all-optical transmission path for its corresponding DB. Consequently the burst is blocked and discarded in an intermediate node. In order to reduce the burst loss probability, many approaches were considered based on different concepts, such as the use of deflection routing to resolve contention presented by Hsu et al. (2002) and Kim et al. (2002). Other promising techniques for partial burst dropping (that reduces the packet loss probability) were introduced, based on the concept of burst segmentation. Optical Composite Burst Switching (OCBS) proposed by Detti et al. (2002), suggests that if all the resources are occupied at the time of the burst arrival, then only the initial part of the burst is dropped. The final part of the burst is transmitted once the needed recourses become available. Similarly, based on the concept of burst segmentation another technique was proposed by Vokkarane et al. (2002) to reduce the packet loss probability. In this technique, which is designed upon Just-Enough-Time (JET) architecture (Yoo and Qiao, 1997), the data burst is broken into multiple segments that consist of a single packet or multiple packets. Combined with deflection routing, the authors have showed that their approach performed better than the entire-burst-dropping policy used by the standard OBS.

Image for - On the Resource Allocation in Optical Burst Switching Networks
Fig. 1: Illustration of data burst contention and segments dropping process

If the burst segmentation strategy is used, the control packets should carry the segments number, their length and possibly their QoS requirements, besides the generic information, e.g., the offset time and the routing information.

In present study, a network resource allocation algorithm is developed for OBS core switches that are implementing burst segmentation techniques to resolve burst contention. With this algorithm the dropped segments are selected evenly from both contending bursts. Furthermore, the technique ensures optimal link utilization and avoids congestion in the control channels by allowing only the bursts with the proper size to be switched and transmitted. The usefulness of the algorithm in the implementation of services differentiation is shown. The performance of the algorithm as well as the services differentiation mechanism is evaluated using a simulation model.

Network resource allocation: A key parameter in the design of an OBS network is the maximum and minimum burst size (Xiong et al., 2000), which is managed by the edge nodes using the assembly algorithms. The minimum length of the data burst is shaped mainly by the electronic processing speed, the switch fabric configuration speed and the ratio of the control channels to the number of data channels in the fiber. To realize an efficient OBS implementation and achieve high link utilization, the data burst transmission time, i.e., burst length/channel speed, should be larger than the switch fabric configuration time. The length of the data burst is entirely overlooked by the resource allocation schemes based on the burst segmentation concept, since no policies related to the size of the truncated burst (i.e., shortened data burst) are implemented, during the burst segmentation process in the core nodes. Furthermore, there is no fairness in allocating the network resources to the contending data bursts, as all the segments are simply discarded from only one burst to resolve the contention, which will increase the likelihood of having bursts shorter than the Minimum Burst Length (MBL).

A better solution would be selecting evenly (fairly) the segments to be dropped from both contending data bursts (whenever possible). Likewise, the truncated burst size should be monitored at the core nodes and guaranteed to be larger than the MBL, which is the minimum length allowed into the network to avoid congestion in the control channels as well as the inefficient bandwidth utilisation.

Besides fairness and data burst size, the technique proposed in this paper is designed to deal with the Switching Time (ST) as a variable that may change from a core node to another. The ST is the time needed to configure the switching fabric (i.e., to switch an output port from one DB to another). To understand what follows, the following definition are provided and illustrated in Fig. 1a.

DBO : Original data burst with arrival time TOA and leaving time TOL.
DBC : Contending data burst with arrival time TCA and leaving time TCL.
TDB : Truncated data burst (i.e., a DB with dropped segments).
N, M : Are respectively the number of segments in DBO, DBC.
DS : Data segment with length DSL.
R : The expected number of segments to be dropped from each data burst,

Image for - On the Resource Allocation in Optical Burst Switching Networks

Before the DBs are sent to the downstream nodes the technique performs three functions arranged in three main events. Starting with Contention_Detection event, if a contention is detected then R (number of segments to be dropped from the contending data bursts) is calculated, then the second event is executed. The second event named Length_of_Truncated_Burst is executed to guarantee that whatever is left from the data bursts after dropping some of their segments is good for transmission over the OBS network. In this event, if one of the truncated data bursts does not meet the MBL requirements, then the contention is simply resolved by dropping the shortest data burst in its entirety. However, the third event is executed if the truncated data bursts are larger than MBL. The Even_Resource_Allocation is used to resolve the burst contention by discarding the overlapping segments alternatively, starting from the tail of the original burst and then the head of the contending burst as shown in Fig. 1a.

The formal description of the technique is presented below.

Image for - On the Resource Allocation in Optical Burst Switching Networks
Image for - On the Resource Allocation in Optical Burst Switching Networks

SERVICES DIFFERENTIATION

Observing the algorithm’s operation, it is apparent that the segments located at the bursts’ ends (tail/head) are more susceptible to be dropped. Therefore, it is proposed to assemble the high priority packets in the segments located at the middle of the burst, i.e., the data segments are assembled into data bursts; in such a way that the lower priority data segments envelop the higher priority data segments as shown in Fig. 1b.

To realize the proposed services-differentiation scheme, both the edge nodes and core nodes must cooperate. The edge nodes should use an appropriate assembly algorithm that complements the aforementioned resource allocation algorithm deployed in core nodes.

Assembly algorithm: Though that either timer-based or burstlenght-based assembly algorithms could be used, a hybrid assembly algorithm is adopted in this scheme. The proposed hybrid approach is accomplished in two stages. First, a burst-length-more precisely segment-length-threshold is used in assembling the data segments from packets with the same QoS requirements (same priority) that will be the segment priority. Second, in constructing the data bursts (from data segments) a timer (time-threshold) is used, that is, after a fixed time, all the data segments (could be with different priorities) destined to the same egress are assembled into a burst, where an optimal maximum burst length could be imposed. The first stage allows the control of the data segments size that must be fixed and restricted to a minimum length. Data padding could be used. For simplicity and to avoid reassembly procedures for IP packets, the size of a segment is set to be large enough to contain the largest IP packet (65,535 B). In the second stage, the use of time-threshold provides uniform intervals between consecutive bursts from the same ingress node into the core networks, besides, managing the packets delay at the ingress nodes. The burst assembly is performed in a way that the lower priority segments will envelop higher priority segments, as illustrated in Fig. 1b.

PERFORMANCE EVALUATION

The performance evaluation of the proposed algorithm is performed using a simulation model and a simple analytical model.

Simulation model: Using NCTUns 2.0 (Wang et al., 2003) the simulation model is developed. As shown in Fig. 2, the simulation topology consists of six traffic sources (hosts 1, 2, 3, 4, 5 and 6), one destination (host 7), two ingress nodes (routers 9 and 10); one egress node (router 11) and one core node (OBS switch 8). IP traffic is sent to the ingress nodes to be aggregated into DBs and then forwarded to the core node. The input and output ports of the OBS switch are monitored. The values of interest are the received data from each input port and the percentage of the forwarded data, as well as the percentage of the DBs with a length less than the MBL.

The MBL is set to be 64 kB and the bursts are equally generated from both ingress nodes. As shown in Fig. 3, the data originating from both ingress nodes is forwarded approximately at the same percentage, if the proposed algorithm is implemented in the core switch (chart a1), otherwise, the amount of forwarded data from router 9 and router 10 are apparently different (chart b1). If the proposed algorithm is implemented, data bursts with length less than the MBL are entirely eliminated (chart a2). However, if a traditional OBS core node (without the proposed algorithm) is used, it is clear that the number of data bursts with length less than the MBL is directly proportional to the traffic load. The values in charts a1 and b1 are based on a single simulation run (to avoid that the results of many simulation runs will counterbalance each other); however, each value in charts a2 and b2 is averaged over five simulation runs.

For the services-differentiation scheme, the simulation model is constructed without wavelength conversion and optical buffer. Only two traffic classes are assumed, with the same offered traffic load. First traffic class with Priority 1 is the higher priority, whereas Priority 0 is for the second traffic class, which has the lower priority. Figure 4 shows that the lower priority data segments (therefore lower priority packets) experience more dropping rate. The reduction on the dropping rate is clear for packets with priority 1.

Image for - On the Resource Allocation in Optical Burst Switching Networks
Fig. 2: Simulation topology

Analytical model: OBS system was modeled as M/M/k/k, M/M/k/D or M/G/n loss system (Yoo et al., 2000; Dolzer and Gauger, 2001; Fan et al., 2002; Dolzer et al., 2001; Liu and Liu, 2002). The well-known Erlang B formula (1) was used to obtain the burst loss probability.

Image for - On the Resource Allocation in Optical Burst Switching Networks
(1)

Where:
A = The traffic load.
k = The number of wavelengths available at each output port.

M/G/k/k queuing system was extended in the literature (Neuts et al., 2002) to an M/G/∞ queue with an unlimited number of pseudo-servers (channels). This model was used to study the performance enhancement in OCBS. Assuming the segment size to be one packet (Vokkarane et al., 2002) and the packets to be fixed in size, we could straightforwardly use the queuing model M/G/∞ to evaluate the performance of the proposed algorithm implemented upon the burst segmentation model proposed by Vokkarane et al. (2002).

With no buffering and seeing that the ST is contained in the dropped DSs, the algorithm can then be modeled as M/G/∞ queue system with infinity of imaginary servers besides the available n servers (i.e., number of wavelengths). With the number of busy servers equal to (n + k), two cases emerge:

k ≤ 0: no contention (number of busy servers is ≤ n).
k > 0: all the n servers are busy and there are k imaginary active servers for k upcoming DBs attempting to be switched.

Under case 2, there are k DSs lost for every n DSs transmitted (for every n DSs on the n servers (wavelengths), k DSs are dropped to resolve the contention).

Image for - On the Resource Allocation in Optical Burst Switching Networks
Fig. 3: Simulation results: Forwarded data vs. Load (a1, b1) and Burst size vs. Load (a2, b2)
Image for - On the Resource Allocation in Optical Burst Switching Networks
Fig. 4: Packet drop rate vs. traffic load for two traffic classes in OBS with and without the proposed algorithm

As soon as the contention is resolved, the contending burst is moved from the imaginary server to be served by an original server. The packet loss probability can be obtained by:

Image for - On the Resource Allocation in Optical Burst Switching Networks
(2)

Where, A and P (n+ k) are respectively the traffic load and the probability that (n + k) servers are busy. Since the number of busy servers in M/G/∞ model has a Poisson distribution (Bertsekas and Gallager, 1992), P(n + k) can be obtained as follows:

Image for - On the Resource Allocation in Optical Burst Switching Networks
Fig. 5: Packet loss probability versus normalized Traffic load

Image for - On the Resource Allocation in Optical Burst Switching Networks
(3)

The results in Fig. 5 show significant performance improvement in the burst segmentation model over the traditional OBS. Further improvement is expected if the OBS system is designed with fiber delay lines, or to support deflection routing.

CONCLUSIONS

In present study, some insight views on the amount of the forwarded data (therefore dropped data) in OBS network are shown; more importantly, the size of the data bursts traveling in the backbone network is investigated. Because the size of the data bursts have a direct impact on the OBS control channels and the bandwidth utilization, a new resource allocation technique has been introduced. The technique is used by the core nodes to monitor and manage the size (length) of the data bursts within the network backbone. Furthermore, this technique achieves better fairness between traffic flows and reduces the probability of having bursts shorter than the MBL, by distributing the dropped segments evenly between the contending bursts. Additionally, the usefulness of algorithm in providing services-differentiation is demonstrated. The technique is simple, practical and its implementation does not lead to any compromises on any of the main drivers behind the emergence of the OBS paradigm, particularly, the design simplicity.

REFERENCES
1:  Battestilli, T., 2002. Optical burst switching: A survey. NCSU Technical Report, 07/2002.

2:  Bertsekas, D. and R. Gallager, 1992. Data Networks. 2nd Edn., Prentice Hall, Englewood Cliffs, New Jersey, pp: 510.

3:  Blumenthal, D., P. Prucnal and J. Sauer, 1994. Photonic packet switches-architectures and experimental implementations. IEEE Proc., 82: 1650-1667.
Direct Link  |  

4:  Chlamtac, I., A. Ganz and G. Karmi, 1992. Lightpath communications: An approach to high-bandwidth optical WANs. IEEE Trans. Commun., 40: 1171-1182.
CrossRef  |  Direct Link  |  

5:  Cruz, R.L. and J.T. Tsai, 1996. COD: Alternative architectures for high speed packet switching. IEEE/ACM Trans. Network., 4: 11-21.
CrossRef  |  Direct Link  |  

6:  Detti, A., V. Eramo and M. Listanti, 2002. Performance evaluation of new technique for IP support in a WDM optical network: Optical composite burst switching (OCBS). IEEE J. Lightwave Technol., 20: 154-165.
Direct Link  |  

7:  Dolzer, K. and C. Gauger, 2001. On burst assembly in optical burst switching networks. A performance evaluation of just-enough-time. Proceedings of the 17th International Teletraffic Congress, September 24-28, 2001, Salvador, Brazil, pp: 149-160.

8:  Dolzer, K., C. Gauger, J. Spath and S. Bodamer, 2001. Evaluation of reservation mechanisms in optical burst switching networks. AEU Int. J. Electron. Commun., 55: 18-26.
Direct Link  |  

9:  Fan, P., C. Feng, Y. Wang and N. Ge, 2002. Investigation of the time-offset-base QoS support with optical burst switching in WDM networks. IEEE Int. Conf. Commun., 5: 2682-2686.
Direct Link  |  

10:  Hsu, C.F., T.L. Liu and N.F. Huang, 2002. On the deflection routing in QoS supported in WDM Optical burst-switched networks. IEEE Int. Conf. Commun., 5: 2786-2790.
Direct Link  |  

11:  Kim, S., N. Kim and M. Kang, 2002. Contention resolution for optical burst switching networks using alternative routing. IEEE Int. Conf. Commun., 5: 2678-2681.
Direct Link  |  

12:  Liu, D.Q. and M.T. Liu, 2002. Differentiated services and scheduling scheme in optical burst-switched WDM networks. Proceedings of the 10th International Conference on Networks, August 27-30, 2002, IEEE Xplore, pp: 23-27.

13:  Mei, Y. and C. Qiao, 1997. Efficient distributed control protocols for WDM optical networks. Proceedings of the International Conference on Computer Communication and Networks, September 22-25, 1997, Las Vegas, NV., USA., pp: 150-153.

14:  Neuts, M., H. Vu and M. Zukerman, 2002. Performance analysis of optical composite burst switching. IEEE Commun. Lett., 6: 346-348.
Direct Link  |  

15:  Qiao, C. and M. Yoo, 1999. Optical Burst Switching (OBS): A new paradigm for an optical internet. J. High Speed Networks, 8: 69-84.
Direct Link  |  

16:  Vincent, W.S., C. Katherine, L. Hall, E. Modiano and A. Kristin, 1998. Rauschenbach. Architectures and technologies for high-speed optical data networks. J. Lightwave Technol., 16: 2146-2168.
Direct Link  |  

17:  Vokkarane, V., J. Jue and S. Sitaraman, 2002. Burst segmentation: An approach for reducing packet loss in optical burst switched networks. IEEE Int. Conf. Commun., 5: 2673-2677.
Direct Link  |  

18:  Wang, S.Y., C.L. Chou, C.H. Huang, C.C. Hwang, Z.M. Yang, C.C. Chiou and C.C. Lin, 2003. The design and implementation of the NCTUns 1.0 network simulator. Comput. Networks, 42: 175-197.
CrossRef  |  Direct Link  |  

19:  Xiong, Y., M. Vandenhoute and H. Cankaya, 2000. Control architecture in optical burst-switched WDM networks. IEEE J. Selected Areas Commun., 18: 1838-1851.
Direct Link  |  

20:  Yoo, M. and C. Qiao, 1997. Just Enough Time (JET): A high speed protocol for bursty traffic in optical networks. Proceeding of the Conference on Technologies For a Global Information Infrastructure, August 31-September 5, 1997, IEEE Xplore, pp: 26-27.

21:  Yoo, M., C. Qiao and S. Dixit, 2000. QoS performance of optical burst switching in IP-Over-WDM networks. IEEE J. Selected Areas Commun., 18: 2062-2071.
Direct Link  |  

22:  Yuan, X., R. Melhem and R. Gupta, 1997. Distributed path reservation algorithms for multiplexed all-optical interconnection networks. Proceedings of the 9th International Symposium on High Performance Computer Architecture, February 1-5, 1997, IEEE Xplore, pp: 38-47.

©  2021 Science Alert. All Rights Reserved