Study and Development of Next-Generation Optical Networks

Smart Computing Review, vol. 4, no. 6, December 2014

470

Smart Computing Review

Study and Development of Next-Generation Optical Networks Taras Maksymyuk, Stepan Dumych, Olena Krasko, Mykola Kaidan, and Bohdan Strykhalyuk Lviv Polytechnic National University / Lviv 79013, Ukraine / [email protected], {stiv.dumych, krasko.lena}@gmail.com, [email protected], [email protected] * Corresponding Author: Taras Maksymyuk

Received August 20, 2014; Revised November 21, 2014; Accepted November 27, 2014; Published December 31, 2014

Abstract: Next-generation optical networks are expected to provide tremendous capacity in order to support upcoming traffic increases. Many technologies are currently being developed for optical transport networks in order to increase throughput, improve energy efficiency and simplify network deployment. The most important problem in current optical networks is transmission of Internet protocol (IP) traffic. Regardless of the tremendous throughput with optical fibers, switching nodes still limit overall network performance. Recently, optical burst switching technology has been developed to overcome this problem. Optical burst switching combines the advantages of both circuit switching and packet switching networks and provides good performance in terms of packet data transmission. Even though optical burst switching networks provide a good mechanism for IP traffic transmission, overall performance is still limited because of access networks. Existing passive optical networks based on Ethernet technology are not fully compatible with optical burst switching, which results in bottlenecks on the border between transport and access networks. In this paper, we present a new method of optical wavelength time-division multiple access (OWTDMA) for passive optical networks. The proposed approach can provide outstanding scalability of network resources and can increase throughput of the optical access network. In addition, we propose implementation of OWTDMA in edge nodes of optical burst switching networks to eliminate bottlenecks between transport and access networks. Simulation results prove the advantage of our proposed approach. Keywords: Optical burst switching, optical wavelength time-division multiple access, passive optical network, IP traffic, burst aggregation

Introduction DOI: 10.6029/smartcr.2014.06.005


471

N

owadays, we face rapid growth in the field of information technology. Currently, the number of Internet-connected devices is approaching 25 billion, which is more than three times the world population. As a result, the amount of traffic on the Internet has greatly increased. According to Cisco Systems Inc., the number of Internet-connected devices will exceed 50 billion by 2020, generating more than 2 zettabytes of Internet protocol (IP) traffic per year. Therefore, a lot of the current research is devoted to designing new architectures and transmission methods for fixed and mobile-access networks [1]. In particular, the upcoming 5G mobile networks are expected to have a 1000-fold capacity compared to the current LTE-A networks [2]. Increasing the total amount of traffic in access networks will increase traffic in transport networks even more. Although current optical networks can provide tremendous capacity by using dense wavelength division multiplexing (DWDM) technology, their performance is limited due to bottlenecks in switching nodes. Wavelength switching technologies can provide good performance when traffic intensity is uniform. However, with rapidly variable IP traffic, the performance of wavelength switching degrades significantly [3]. Recently, packet switching technologies have been developed to overtake the threshold of circuit switching networks. These technologies are summarized under a new paradigm called IP over DWDM (IPoDWDM) [4]. Packet switching networks can handle abrupt changes in traffic intensity, but this results in higher overhead, and decreases network performance. The most promising and scalable IPoDWDM implementation is the optical burst switching (OBS) network [5]. OBS combines the advantages of both circuit switching and packet switching by providing a scalable architecture for optical transport networks and flexible traffic management [6]. Even though throughput of an OBS transport network is theoretically unlimited, we still face limitations in edge node performance in terms of traffic aggregation from access networks to transport networks, and vice versa. In this paper, we introduce a new type of optical access network—optical wavelength time division multiple access (OWTDMA). In addition, a flexible burst aggregation method for OBS networks eliminates bottlenecks in edge nodes and increases overall network performance. The rest of this paper is organized as follows. Section 2 briefly introduces optical burst switching transport networks. Section 3 describes a detailed study of passive optical networks and the proposed orthogonal frequency division multiple access (OFDMA) network, which includes a traffic aggregation process from separate terminal nodes. Section 4 presents the simulation and the performance analysis of optical burst switching and wavelength time-division multiple access. Section 5 concludes the paper.

Architecture and Data Transmission Process in Optical Burst Switching Networks Circuit switching optical networks provide direct continuous channels between nodes, separated by different wavelengths. In this case, channels are constantly utilized for transmission sessions, providing good throughput. However, when traffic intensity is very low, throughput of optical channels is underutilized, resulting in lower network capacity [7]. Packet switching optical transport networks allow an increase in network performance because throughput of each channel is utilized more effectively by using multi-protocol label switching (MPLS) [8]. However, for high-intensity traffic, packet switched networks generate high overhead from signaling data to support transmission of each packet. In optical burst switching networks, all IP packets are assembled into logical bursts and transmitted all at once. Burst is a logical combination of homogeneous IP packets with the same destination node and similar quality of service (QoS) requirements. Bursts are created in order to rearrange traffic from different sources and separate traffic according to throughput requirements and destination address [9]. In order to coordinate burst transmission, a burst header packet (BHP) is created for each burst. The BHP contains information about source and destination OBS nodes, QoS requirements and additional control data to support burst transmission, scheduling and switching, which can vary depending on the selected signaling scheme in an OBS network [9]. Optical burst switching transport networks are usually based on a mesh topology, where each node can be connected to multiple neighbor nodes. Each optical switch combines two parts separated according to their functionality: edge nodes and core nodes [9]. An edge node is responsible for traffic aggregation from multiple access networks, creating bursts of IP packets and generating a BHP for them, or vice versa. A core node is responsible for providing burst routing and switching based on information obtained from the BHP. Burst transmission and switching is carried out in transparent fashion, without data conversion to electric representation, by using all-optical switching technologies. In order to provide this, BHP transmission is scheduled ahead of bursts, and creates a transparent virtual channel to the destination node [10]. Because all optical switching time is comparable to signal propagation time through optical fibers, for higher layers, burst transmission between two edge nodes is logically considered a direct connection, even though each burst really transmits through many hops. An example of data transmission in an OBS network through the virtual channel is shown in Figure 1. As mentioned above, the optical switch combines edge node and core node. But in Figure 1, core and edge nodes are presented as different switches for better understanding.

Maksymyuk et al.: A Study and Development of the Next Generation Optical Network

472

IP access network

IP access network

OBS network

• Burst creation • BHP generation • Sending BHPs and bursts

Optical channel

• BHP is transmitted ahead of burst, having enough time to “open” virtual channel • Burst is transmitted quickly without delay in each node, catching BHP in destination node

Core nodes

IP access network

IP access network

Virtual channel

Edge node A

Edge node B

• BHP analysis • IP routing • Burst all-optical switching

• BHP analysis • Burst reception • Scheduling and distributing of IP packets in access network

IP access network Figure 1. Data transmission process in optical burst switching networks

■ Edge Node Architecture and Functionalities The edge node is a very important part of an OBS network and provides intermediate functions between transport and access networks. IP traffic is aggregated from many different access networks, such as passive optical networks (PON), enterprise networks, cellular networks, cloud data center networks, etc. As a result, incoming traffic at the edge node is highly heterogeneous, because each access network sets different quality of service requirements, such as throughput, latency, and packet loss probability. For example, packets from a cellular network may have strict requirements for latency, because they have already been delayed due to processing in the cellular network system. But packets from an optical access network with similar initial service demands can be delayed longer in an edge node, providing the same experience for end users. On the other hand, non–real time packets with high throughput demands can stay for a long time but utilize higher throughput for transmission (e.g. online video-on-demand services). In addition, an OBS network should support transmission of IPv6 traffic, which differs from IPv4 in terms of packet structure. That brings compatibility problems, in addition to current QoS requirements. Therefore, traffic aggregation and burst scheduling in the edge node are very complicated tasks because of variable parameters, and there are still no optimal solutions for burst creation and transmission problems. Figure 2 presents the burst aggregation process in the edge node of an optical burst switching network [11]. First, incoming streams of IP packets from different segments of access networks are multiplexed into one stream with respect to arrival time. Second, the packets’ classifier distributes packets into corresponding bursts and reports burst status to the scheduler. Simultaneously, the scheduler assigns the necessary output fiber and the corresponding wavelength for each burst, reporting to the BHP generator. Third, the BHP generator creates BHP packets according to the data obtained from the scheduler. Fourth, the scheduler sends the BHP and then sends the corresponding burst after a predefined delay, called the offset time. Offset time is predefined for each route to ensure that the BHP is successfully processed in the destination node before the corresponding burst reaches the node. Because the burst transmits without processing in core nodes, a “chasing” effect is observed where the distance between the BHP and the corresponding burst decreases after each intermediate core node. In this case, a virtual channel will be active for the exact burst transmission, utilizing throughput most effectively. One of the important parts in the burst aggregation process is determining the optimal size for each burst. By size, we mean the total of all IP packets collected into a burst. Considering the offset time between BHP and burst, it is obvious that


473

larger bursts provide better throughput utilization compared to smaller bursts because fewer bursts are required to transmit the same amount of data. However, maximal burst size is also limited because, for low traffic–intensity transmission, the deadline of the first packet in the burst can arrive before the burst is full. Therefore, the size threshold will not be optimal for burst transmission. Another approach is to create a time threshold for burst transmission according to the first packet’s deadline. But this results in buffer overload when traffic intensity increases, because the burst’s size can be too large before the deadline arrives. Therefore, an adaptive algorithm was applied by Dumych et al. [12] to provide a tradeoff between delay time and burst size depending on traffic intensity.

Burst scheduler

BHP generator

1 IP stream IP stream

…

IP stream

1 Optical channel

MUX

Packet classifier

2

…

N

2

x

x

N

Burst/BHP buffer Figure 2. Burst aggregation process in the edge node of an optical burst switching network

■ Core Node Architecture and Functionalities The core node in an OBS network provides all-optical switching for each incoming burst to the target destination node. A core node consists of two parallel systems: an all-optical switching system and a BHP processing system [13]. The BHP processing system analyzes incoming BHPs in order to get information about burst destination nodes and QoS requirements, burst size and arrival time. According to the obtained information, the BHP processor provides complex routing by taking into account all incoming bursts in the same time interval, allowing distribution to available channel resources with respect to QoS requirements. The all-optical switching system provides a burst switching process without optical-to-electrical and electrical-to-optical conversion, according to the obtained switching matrix from the BHP processing system. The alloptical switch can be based on different optical switching technologies, such as micro-electro-mechanical switching, electro- and acousto-optical switching, or liquid crystal switching [14,15]. The optical burst switching process in a core node is shown in Figure 3. Incoming bursts and BHP packets are transmitted via DWDM channel with n wavelengths. A demultiplexer (DMUX) separates all incoming wavelength channels and divides them between BHP channels and burst channels. As shown in Figure 3, wavelengths from λ1 to λx are used for transparent burst transmission, and these channels are connected directly to the all-optical switch. The rest of the channels with wavelengths from λx+1 to λn are connected to the BHP processing system and are responsible for BHP transmission. The switching process in the core node is described below. First, BHPs arrive in the BHP processing system. BHPs are converted from optical to electrical form, which allows routing, scheduling and switching by means of digital processing devices. Second, the BHP analyzer carries out the scheduling process according to the bursts’ arrival times and classifies bursts between corresponding time slots. Third, a router provides a complex routing algorithm to compute optimal channel allocation with respect to QoS demands of all upcoming bursts within the current time slot. After that, the switching controller rearranges the optical switch according to information from the router. Finally, the BHP generator creates updated BHPs, converts them to optical form, and transmits them to the next-hop neighbor nodes. Eventually, bursts will arrive at the edge node and switch to the destination without conversion to electrical form.

Optical Wavelength Time-Division Multiple Access for Optical Burst Switching Networks


474

BHP processing system

O/E

E/O

λn

λn

Router

…

BHP analyzer

…

BHP generator

O/E

λx … λ3 λ2 λ1

…

DMUX

λx+1

Switching controller

Burst Header Packet (BHP)

λ1....λn

O/E – optical-to-electrical converter E/O – electrical-to-optical converter

E/O

λx+1

Optical burst

All-optical switching system

λx … λ3 λ2 λ1

…

MUX

λ1....λn

Figure 3. Burst switching process in the core node of an OBS network As mentioned in the previous section, data aggregation and disaggregation in the edge node of an OBS network significantly influences overall network performance. Many studies of this problem have been done in order to improve the burst aggregation process by using burst size optimization and adaptive traffic classification. However, the main limitation for edge node performance is access networks that were designed before OBS technology and which are thus are not fully able to utilize all the advantages of optical burst switching. The most common access network for OBS is the gigabit Ethernet passive optical network (GEPON) [16]. In a GEPON, all packets from different sources are transmitted by using time-division multiple access (TDMA), which can partially adjust throughput between different users according to demand. However, this results in a synchronization problem between the optical line terminal (OLT) and the optical network unit (ONU). In addition, it results in undetermined packet arrivals at the edge node, which significantly increases the complexity of the burst aggregation and scheduling problem. ONU response time

OLT

GATE 1

…

GATE N

REPORT 1

Downlink delay

ONU

GATE 1

…

…

REPORT N

t

Uplink delay

GATE N

REPORT 1

…

REPORT N

t

ONU delay

Figure 4. MPCP protocol diagram of OLT and ONU interaction

■ Data Transmission and Multiplexing in GEPONs Interaction between the OLT and multiple ONUs is supported by multi-point control protocol (MPCP) management [17].


475

MPCP is based on two types of control message: GATE and REPORT. The GATE message is sent from the OLT to the ONUs to provide them with necessary information about upcoming transmissions, such as communication mode, receiver identification, corresponding time slots for downlink and uplink transmissions, etc. Each ONU responds with a REPORT message, providing the information about its condition. The OLT needs the REPORT messages from the ONU in order to distribute the available throughput between each node in the uplink channel. MPCP operates in one of two modes: normal mode and initialization mode. Normal mode is applied for the usual transmission of control messages between the OLT and ONU in order to support the communications process in both downlink and uplink channels. Initialization mode is applied from time to time to detect a new ONT and its registration by the OLT for further recognition and involvement in the communications process. In downlink transmission, the OLT broadcasts TDMA frames to end users. Each ONU selects its time slot, which was predefined by the MPCP. On the uplink channel, users perform simultaneous frame transmission. Each ONU allocates its own fixed time slot, which was also predefined by the MPCP, and transmits user data during this time slot. Even if the user data rate is less than the available throughput provided by the time slot, the time slot duration remains constant by complementing the user's data with idle segments to maintain synchronization. This results in a significant decrease in throughput utilization. The downlink and uplink data transmission process for a GEPON is shown in Figure 5.

ONU 1

1

OLT

2

1

3

1

2

1

3

1

1

USER 1

2

ONU 2

USER 2

802.3 frame Header

DATA

FCS

3

ONU 3

USER 3

(a)

ONU 1

2

OLT

1

1

2

3

ONU 2

1

1

USER 1

2

USER 2

Time slot 802.3 frame

Header

DATA

FCS

ONU 3

3

USER 3

(b) Figure 5. (a) Downlink and (b) uplink data transmission in a passive optical network

■ Optical Wavelength Time-Division Multiple Access Network We propose a new optical wavelength time-division multiple access (OWTDMA) method for passive optical networks to


476

mitigate the synchronization problem between OLTs and ONUs. We developed an OWTDMA resource grid for precise throughput allocation. The resource grid consists of time slots and wavelengths similar to OFDMA [18]. The difference with our approach is that resources are distributed amongst separate carriers, while the OFDMA resource grid is allocated to one carrier band divided into narrow subcarriers. Thus, OWTDMA provides the full bitrate per carrier, while the bitrates of OFDMA subcarriers are n times lower, where n is the number of subcarriers. Figure 6 shows the data transmission process in an OWTDMA passive optical network. Broadcast channels

λ

t REPORT messages

Header

ONU 1

USER 1

ONU 2

USER 2

ONU 3

USER 3

FCS

n

…

2

1

OLT

…

2

n

1 n

1

…

2

2

… 1

n

GATE messages

Header

Resource element

FCS

t λ

Broadcast channels

Figure 6. Downlink and uplink data transmission in an OWTDMA passive optical network As shown in Figure 6, in an OWTDMA passive optical network, uplink and downlink transmission are provided by the same resource grid. The difference is only in the used wavelengths in order to separate uplink and downlink channels in a single optical fiber. Utilizing the OWTDMA resource grid allows better resource scalability by allocating the necessary number of resource elements for each end user. Moreover, OWTDMA allocates the necessary number of broadcast channels for transmission, and supports all MPCP functionalities. The advantage of our proposed approach, compared to an ordinary gigabit Ethernet network, is that we mitigate the synchronization problem. Instead of adjusting throughput between users by changing channel occupation time, we provide a fixed time frame for OWTDMA resource grid transmission. Throughput is adjusted by allocating the necessary number of resource elements for the user’s data. Therefore, instead of defining corresponding time slots, the MPCP provides resource mapping for each ONU. By knowing its resource mapping, each ONU will access its resource elements directly, providing users with the necessary data. In addition, the proposed resource allocation increases performance of the uplink channel, because there are no underutilized time slots. This allows effective allocation of the available throughput, even if traffic intensity is low.

■ OWTDMA Burst Aggregation Method in Edge Nodes of an OBS Network In order to improve performance of optical burst switching, we propose an OWTDMA-based burst aggregation method in edge nodes. We exploit MPCP functionality in the optical burst switching network. Because MPCP is already familiar with user mapping in the OWTDMA resource grid, we can share this information with the edge node of an OBS network. Therefore, the burst aggregation process can be simpler because of direct association between bursts and the corresponding OWTDMA resource elements. In this case, we can increase burst size during the same aggregation time because of parallel packet transmission. OWTDMA also increases performance of burst disaggregation, when traffic streams with different QoS demands can be transmitted simultaneously with a different number of allocated OWTDMA resource elements. The burst aggregation process in an edge node for the OWTDMA method is presented in Figure 7.


Burst scheduler

477

BHP generator

Header

1

1

2

2

Optical channel

MUX

…

N λ

t

x

x

N

Burst/BHP buffer

Figure 7. OWTDMA burst aggregation method in an edge node of an OBS network As shown in Figure7, incoming OWTDMA resource blocks consist of data resource elements and header resource elements. To simplify the packet classification process and the QoS requirement analysis, OWTDMA just copies data from header resource elements to the burst scheduler. This eliminates the complicated traffic classification process, because all necessary resource mapping data is already known from the OLT of the OWTDMA passive optical network. We also assume that separate IP streams from different ONUs can have the same destination node and similar QoS demands. In that case, the burst scheduler combines the two resource maps into one and starts to treat multiple IP streams as one. This simplifies burst scheduling and increases performance in the OBS network.

Simulation and Performance Analysis of Optical Burst Switching and Wavelength Time-Division Multiple Access We conducted a simulation of the burst aggregation process in the edge node of an optical burst switching network. We assumed that incoming traffic in the OBS network is self-similar, as was proven by Ge et al. [19]. Therefore, we simulated an edge node as a queueing system G/M/1 with normal distribution of incoming traffic and Poisson distribution of IP packet size. We studied the performance of three existing burst aggregation methods: buffer threshold, time threshold and our previously proposed adaptive threshold method [12] based on the detection of an abrupt buffer load increase. Simulation results are shown in Figure 8.

(a) (b) Figure 8. Simulation results of burst aggregation methods: (a) buffer load, and (b) burst size distribution

478


As seen from the obtained results, the buffer threshold method results in the smallest burst size distributed around 100 kB (Figure 8b). However, during two hours of simulation, the buffer load for the buffer threshold algorithm did not reach higher than 60% (Figure 8a). For the time threshold method, bursts aggregated to a much larger size, distributed around 250 kB (Figure 8b). But the buffer overloaded frequently, which resulted in a 30% packet loss. The adaptive threshold method for burst aggregation was proven an effective solution for OBS networks. According to our simulation, the buffer always loaded up to 90%, but almost never touched the 100% limit, resulting in 3% lost packets (Figure 8a). Despite this, burst size was distributed around 200 kB (Figure 8b), which seems to be large enough for effective throughput utilization, but still less than the time threshold method. We also studied the performance of the proposed optical wavelength time-division multiple access in passive optical networks. We provided a simulation of a passive optical network for ordinary gigabit Ethernet and for OWTDMA. To ensure a fair comparison, we limited total throughput to 10 Gbps, even though OWTDMA can provide much more thanks to DWDM technology. We simulated the data transmission process on uplink (Figure 9a) and downlink (Figure 9b) for two hours in order to study the pure advantage of our proposed approach.

(a) (b) Figure 9. Comparison of throughput between GEPON and OWTDMA PON for (a) an uplink channel and (b) a downlink channel As shown in Figure 9, optical wavelength time-division multiple access provides better throughput compared to the gigabit Ethernet PON. However, for the downlink channel, the performance gain is not sufficient. OWTDMA provided only a 3% average throughput gain, compared with GEPON. On uplink, OWTDMA provided a 13% gain on average throughput, compared with GEPON. This gain was achieved by eliminating the idle segments in OWTDMA during uplink transmission by using a wavelength-time resources grid.

Conclusions This article provided a comprehensive study of data transmission in optical transport and access networks. We addressed the most important issues in optical burst switching networks, such as the burst aggregation problem and optimal burst size. We simulated performance of the burst aggregation process in edge nodes of an OBS network. From the obtained results, we can see that the adaptive algorithm is optimal for achieving good tradeoff between throughput and packet loss. In addition, QoS requirements should be considered as well, in terms of packet loss and latency requirements. It is likely that performance of the burst aggregation process will vary depending on many additional circumstances. In our future work, we will provide a more comprehensive study of the traffic aggregation process in OBS networks. We further developed this area by introducing a new method for optical access networks, called OWTDMA, which provides good flexibility in resource allocation and improves performance of a passive optical network. Simulation results proved the advantage of our approach, especially on uplink. We also extended OWTDMA to the edge node of an OBS transport network to improve the burst aggregation/disaggregation process. We propose a new method of burst aggregation in optical burst switching networks by using the OWTDMA resource frame combined with header packets. From separate processing of data and header packets, we eliminated the complicated process of IP packet classification, which allows


479

simplification of burst aggregation and BHP generation in edge nodes. In future work, we will try to integrate OBS and OWTDMA more closely in order to develop the next generation of optical network infrastructure.

References [1] G. Shen, R. Tucker, “Fixed mobile convergence (FMC) architectures for broadband access: Integration of EPON and WiMAX,” Asia-Pacific Optical Communications, International Society for Optics and Photonics, pp. 678403-678403, Nov. 2007. [2] N. Bhushan, J. Li, D. Malladi, R. Gilmore, D. Brenner, A. Damnjanovic, R. T. Sukhavasi, C. Patel and S. Geirhofer, “Network densification: the dominant theme for wireless evolution into 5G,” IEEE Communications Magazine, vol. 52, no.2, pp. 82-89, 2014. Article (CrossRef Link) [3] L. Xu, H.G. Perros, G. Rouskas, “Techniques for optical packet switching and optical burst switching,” IEEE Communications Magazine, vol. 39, no. 1, pp.136-142, 2001. Article (CrossRef Link) [4] S. Dixit. “IP over WDM: building the next-generation optical internet,” John Wiley & Sons, 2004. [5] Y. Chen, C. Qiao, X. Yu, “Optical burst switching: a new area in optical networking research,” IEEE Network, vol. 18, no. 3, pp. 16-23, 2004. Article (CrossRef Link) [6] S. Dumych, P. Guskov, T. Maksymyuk and M. Klymash, “Simulation of characteristics of optical burst switched networks”, in Proc. of IEEE International Conference on Microwave & Telecommunication technology (CriMiCo’2013), pp. 492-493, Sep. 2013. [7] R. Ramaswami, K. N. Sivarajan, “Routing and wavelength assignment in all-optical networks,” IEEE/ACM Transactions on Networking (TON), vol. 3, no. 5, pp. 489-500, 1995. [8] X. Xiao, A. Hannan, B. Bailey, L. M. Ni, “Traffic Engineering with MPLS in the Internet,” IEEE Network, vol. 14, no.2, pp. 28-33. Article (CrossRef Link) [9] K. Dolzer, C. Gauger, “On burst assembly in optical burst switching networks – a performance evaluation of JustEnough-Time,” in Proc. of ITC 18, pp. 149-161, Sep. 2001. [10] S. Dumych, T. Maksymyuk, O. Krasko, and P. Guskov “The Virtual Channel Parameters Calculation in All-Optical Network,” in Proc. of IEEE International Conference on The Experience of Designing and Application of CAD Systems in Microelectronics, (CADSM 2013), p.87, Feb. 2013. [11] V. M. Vokkarane, J. P. Jue, “Prioritized burst segmentation and composite burst-assembly techniques for QoS support in optical burst-switched networks,” IEEE Journal on Selected Areas in Communications, vol. 21, no. 7, pp. 11981209, 2003. Article (CrossRef Link) [12] S. Dumych, T. Maksymyuk and P. Guskov, “Simulation of Burst Aggregation and Signaling schemes for Optical Burst Switched Networks”, in Proc. of Int. Conf. Computer Science and Engineering, Lviv Polytechnic National Univ., Ukraine, pp.40-41 Nov. 2013. [13] L. Liu, et al, “Experimental comparison of high-speed transmission control protocols on a traffic-driven labeled optical burst switching network test bed for grid applications,” Journal of Optical Networking, vol. 8, no. 5, pp. 491503, May 2009. Article (CrossRef Link) [14] T. Krause, “Migration to All-Optical Networks,” Alcatel Network Systems, 1998. [15] A.S. Andrushchak, M.V. Kaidan, Ye. M. Chernyhivskiy, O.V. Yurkevych, T.A. Maksymyuk, B.G. Mytsyk, and A.V. Kityk, “Application efficiency increasing of LiNbO3: MgO and GaP crystals for acoustooptical high-frequency control of powerfull laser irradiation,” in Proc. of IEEE International Conference on Laser and Fiber-Optical Networks Modeling, (LFNM 2010), pp.173-175, 2010. [16] A. Yin, P. Zhang, W. Zhou, Y. Jiao, “Dynamic Programmable Scheduling Mechanism of GEPON,” Journal of Lightwave Technology, vol. 27, no.19, pp. 4289-4296, 2009. Article (CrossRef Link) [17] X. Liu, G.N. Rouskas, F. He and H. Xiong, “Multipoint Control Protocol with Look-Ahead for Wavelength Division Multiplexed Ethernet Passive Optical Network,” Journal of Optical Communications and Networking, vol. 6, no. 2, pp. 104-113, 2014. Article (CrossRef Link) [18] A. J. Lowery, L. B. Du, J. Armstrong, “Performance of optical OFDM in ultralong-haul WDM lightwave systems,” Journal of Lightwave Technology, vol. 25, no.1, pp.131-138. Article (CrossRef Link) [19] A. Ge, F. Callegati, L.S. Tamil, “On optical burst switching and self-similar traffic,” IEEE Communications Letters, vol. 4, no.3, pp. 98-100, 2000. Article (CrossRef Link)

480


Taras Maksymyuk received his MS in information communication networks from Lviv Polytechnic National University, Ukraine, in 2011, and his BA in telecommunications from Lviv Polytechnic National University, Ukraine, in 2010. He has been a PhD student at Lviv Polytechnic National University, Ukraine, since 2012. Currently, he is a researcher with the Department of Computer and Information Science at Korea University, Sejong Metropolitan City, South Korea. His research interests include converged networks, mobile cloud computing, optical transport and access networks, 5G, heterogeneous networks, and software-defined networks.

Stepan Dumych received his BS and MS in radio physics from Ivan Franko National University of Lviv, Ukraine, in 2004 and 2006, respectively. He is currently a senior teacher with the Telecommunications Department of Lviv Polytechnic National University. His research interests include design and synthesis of microprocessor systems, FPGA and optical burst switching.

Olena Krasko received her MS in information communication networks from Lviv Polytechnic National University, Ukraine, in 2008, and her BA in telecommunications from Lviv Polytechnic National University, Ukraine, in 2007. She is currently an assistant with the Telecommunications Department of Lviv Polytechnic National University. Her current research interests include optical access networks, IP over DWDM and software-defined networks.

Mykola Kaidan received his PhD in optoelectronics from Lviv Polytechnic National University, Ukraine, in 2006, and his MS in optoelectronics from Lviv State University named after Ivan Franko in 1999. He is currently in post-doctoral study on telecommunications systems and networks at Lviv Polytechnic National University, Ukraine. He is a docent with the Telecommunications Department of Lviv Polytechnic National University. His current research interests include optical networks, photonic optical fibers, non-linear fiber optics and optical network components.

Bohdan Strykhalyuk received his PhD in telecommunications systems and networks from the Odessa National Academy of Telecommunications named after A.S. Popov, Ukraine, in 2010, and his MS degree in radio engineering from Lviv Polytechnic Institute in 1985. He is currently in post-doctoral study into telecommunications systems and networks at Lviv Polytechnic National University, Ukraine. He is a docent with the Telecommunications Department of Lviv Polytechnic National University. His current research interests include cloud computing, complex networks, network convergence, big data, optical transport networks and service-oriented networks.

Copyrights © 2014 KAIS