Recent Dynamic Bandwidth Allocation Algorithm Methods ... - CiteSeerX

International Journal of New Computer Architectures and their Applications (IJNCAA) 4(4): 167-176 The Society of Digital Information and Wireless Communications, 2014 (ISSN: 2220-9085)

Recent Dynamic Bandwidth Allocation Algorithm Methods in Ethernet Passive Optical Network

a

*Nurul Asyikin Mohd Radzia, Norashidah Md. Dina, Mohammed Hayder Al-Mansoorib, Sajaa Kh Sadona Center for Communication Services Convergence Technologies, College of Engineering, Universiti Tenaga Nasional, Jalan IKRAM-UNITEN, 43000 Kajang, Selangor, Malaysia b Faculty of Engineering, Sohar University, PO Box 44, PCI 311, Sohar, Oman *Corresponding author: [email protected]

ABSTRACT The “last mile” bandwidth bottleneck problem can be best solved by using Ethernet Passive Optical Network (EPON). EPON is a promising technology to implement the Fiber To The Home solution. Dynamic bandwidth allocation (DBA) is one of the critical issues in the design of the EPON system that has been studied by many researches. This paper introduces the fundamental concepts on EPONs, its protocol and its bandwidth allocation. We also compile and classify the research works in order to provide the state-of-art of the DBA algorithms. The advantages and disadvantages of each classification is provided. The classifications provide insightful presentations of the prior work on EPONs which will help the researches to choose the most suitable algorithm as according to the needs.

KEYWORDS EPON, DBA, survey

1 INTRODUCTION Ethernet passive optical network (EPON) is a point-tomultipoint fiber topology that has a nominal rate of 1.25 Gbps. However due to a 8B/10B encoding, its effective rate is 1 Gbps. The maximum distance of an EPON is 20 km. The typical topology used in EPON is tree topology. This is because tree topology offers the most flexibility in adapting to a growing subscriber base and increasing bandwidth demands [1]. An EPON system composes of an optical line terminal (OLT), multiple optical network units (ONUs), an optical distribution network (ODN) and optical fiber. The term “passive” in EPON is used because the only active elements used in the system are OLT and ONU.

OLT is typically an Ethernet switch or media converter platform that is located at the service provider’s premises. It is used not only to transmit an optical signal over a fiber, but also to manage communications to the multiple subscribers. The user’s premises are where the ONUs are located. Its function is to convert the optical signal to electrical signals for use by various devices such as phones, computers, televisions, etc. The ones that connect one OLT to multiple ONUs is called ODN. ODN can either be one or more passive optical splitters that function to split downstream signal from one fiber into multiple fibers and combine upstream signals from multiple fibers into one. The multiple fibers depend on the number of ONUs, which is normally between 4 and 64 as specified in the IEEE 802.3 standard, Clause 60 [2]. EPON can either be performed in downstream transmission or upstream transmission. Downstream transmission happens when OLT transmits frames to ONUs with a wavelength of 1490 nm. In the downstream transmission, frames transmitted from the OLT are tagged with a logical link identifier (LLID) at the preamble of each frame. These frames are then broadcasted to every ONU in the system. The purpose of tagging such frames is to allow each ONU to extract and filter only its own frames because every ONU will eventually receive all frames transmitted from the OLT. On the other hand, the upstream transmission happens when frames are transmitted from the ONUs to the OLT at a wavelength of 1310 nm. Recall back the architecture of EPON is point to multipoint. With upstream transmission, when multiple ONUs transmit frames to the OLT at the same time, the frames need to share the same fiber from the splitter to the OLT. This

167


causes collision that can be avoided using an arbitration mechanism. However, an arbitration mechanism in EPON does not specify any particular allocation algorithm. Therefore in this paper we will review the recent dynamic bandwidth allocation (DBA) algorithms that have been proposed in the literature. The remainder of the paper is as follows. In Section 2 we explain on the arbitration mechanism of EPON. Section 3 discusses on the allocation algorithm methods in EPON. In Section 4 we present a survey of the recent state-of-the-art DBA algorithms for EPON systems. In Section 5, we conclude this paper.

2 MULTIPOINT CONTROL PROTOCOL The arbitration mechanism mentioned in previous section is called multipoint control protocol (MPCP). MPCP supports time slot allocation from the OLT to the ONUs. It provides a framework intended to facilitate the implementation of bandwidth allocation algorithms by providing signaling infrastructure for coordinating upstream data transmission. There are two types of modes in MPCP; Autodiscovery mode and Normal mode. At the beginning of the communication process, an autodiscovery mode is used to register newly connected ONUs. This is done without manual intervention so that ONU can join the EPON system without affecting other ONUs. Sometimes, existing ONUs may lost synchronization with the OLT. Autodiscovery will help to resynchronize these ONUs. It employs four MPCP messages which are discovery GATE, REGISTER_REQ, REGISTER and REGISTER_ACK [2]. Autodiscovery works when OLT transmits a discovery GATE message to an ONU. Discovery GATE message is used to discover the slot and length of a new ONU. Discovery GATE message will first be timestamped with the OLT’s local time. Once an ONU receives the discovery GATE message, it will set its local time to the timestamp received. Then, ONU sends the REGISTER_REQ message to the OLT. The REGISTER_REQ message contains ONU’s source address and timestamp so that the OLT is able to learn the ONU’s media access control (MAC) address and round trip time. Afterwards, the OLT will send the REGISTER message that contains an LLID. This LLID serves as a unique identification value of ONU.

Other than that, it will also send a normal GATE message to that same ONU. ONU will finally send the REGISTER_ACK message as an acknowledgement that the ONU has received the message. After the new ONUs has been discovered, we need to sustain the communication between OLT and ONUs. This is the task of the normal mode. Normal mode employs two MPCP messages namely REPORT and GATE messages. A REPORT message is sent by an ONU to the OLT to request for a time slot whereas a GATE message is sent from the OLT to an ONU to grant the time slot. ONU will first send a REPORT message to the OLT. The REPORT message can either be transmitted automatically or on demand at either the start or the end of a window. The OLT will then send a GATE message to the ONU as a reply. Here, MPCP will timestamp the GATE message with its local time before sending the message. This is to allow the ONU to program its local register and update its local clock as per timestamp extracted from the received GATE message upon receiving it. This is important in order for ONU to maintain synchronization with the OLT. The ONU will start to transmit frames for up to the window size at the granted start time.

3 BANDWIDTH ALLOCATION Since IEEE 802.3ah does not specify any method to allocate the time slot from OLT to the ONUs [2], it allows them to be vendor-specific. The directional properties of the splitter in EPON make the conventional Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) and Carrier Sense Multiple Access with Collision Detection (CSMA/CD) methods difficult to implement in EPON [3]. This is because an ONU is unable to inform other ONUs about the collision. Although OLT is able to detect or avoid a collision and inform the ONUs about the collision by sending a jam signal, propagation delays in EPON which can exceed 20 km in length can greatly reduce the efficiency of the system. Therefore, there is no guarantee that an ONU can get an access to the medium without any delay. Since EPON must support real time traffic such as voice and video traffic, it must be able to guarantee the delivery of these traffic types. Thus, the conventional CSMA/CD is not suitable to be used in EPON. Although CSMA/CA is able to support real time applications in EPON, it is still not efficient to be used in EPON due to its propagation delay. 168


The optical looping-back technique was then proposed by Desai et al. [4] to achieve high channel efficiency with CSMA/CD. With this method, a portion of upstream signal power transmitted by each ONU is looped back to the other ONUs by using a coupler. Two ports of the coupler are connected together via isolator. If two or more ONUs transmit data simultaneously, collisions will be detected at each ONU and all data transmissions will be stopped immediately. However, to implement this technique, each ONU has to use an additional receiver and a carrier sensing circuit. This is not a preferred method since it increases the network cost and it is unable to provide guaranteed bandwidth. Another possible solution is to use wavelength division multiplexing (WDM) method [5]. This method allows each ONU to operate at a different wavelength to avoid collision. In order to receive the data transmitted in multiple channels, a tunable receiver or a receiver array is required at the OLT. It also requires each ONU to use a fixed transmitter operating at a different wavelength, which would result in an inventory problem. Although the inventory problem can be solved by using tunable transmitters, these devices are costly, making the solution cost-ineffective. At this moment, time division multiple access (TDMA) is considered to be the most effective solution for upstream EPON [6]. In TDMA, OLT allocates a time slot or a transmission window for data transmission in ONU. Upon the arrival of its time slot, the ONU will send out its buffered packets at the full transmission rate of the upstream channel. If there are no frames in the buffer to fill the entire time slot, idles are transmitted. TDMA can be either static or dynamic, depending on the arbitration mechanism implemented by the OLT. Static bandwidth allocation (SBA) is simple to implement [7]. With SBA, once bandwidth is assigned to an ONU, it will be unavailable to other ONUs in EPON. Due to the bursty nature of network traffic, it may result in a situation where several time slots overflow even under very light load, causing packet delay for several time slots, while other time slots are not fully utilized even under very heavy traffic. For this reason, the SBA is not preferred. To increase bandwidth utilization, it is desirable that the OLT dynamically allocates a variable time slot to each ONU based on the instantaneous bandwidth

demand of the ONUs. Dynamic bandwidth allocation (DBA) is suitable because the IP traffic is burst traffic. With DBA, when a particular ONU is not using its allocated bandwidth, that bandwidth can be reassigned to another ONU. This feature enables the flexibility of DBA to meet the different type of traffics in EPON. Without DBA, the unused bandwidth would be stranded and unusable by other ONUs on the network.

4 DBA METHODS Figure 1 shows the taxonomy of DBA algorithm that is used as a framework to discuss the research on DBA. The DBA methods are divided into two categories; with quality of service (QoS) support and without QoS support. Dynamic Bandwidth Allocation (DBA)

Do not support QoS

Support QoS

Support QoS locally

Limited

Prediction

Excessive bandwidth

Support QoS universally

Others

Fuzzy Logic

Figure 1. DBA taxonomy

4.1 DBA algorithms without QoS support The first DBA algorithm for EPON that can be found in the literature is Interleaved Polling with Adaptive Cycle Time (IPACT) [8]. In IPACT, the OLT polls ONUs and grants the bandwidth to each ONU in a round-robin fashion according to the ONU’s bandwidth demand. Each ONU is served once per round-robin polling cycle. The length of the polling cycle is not fixed where it adapts to the bandwidth requirements of the ONUs. The dynamic cycle length may result in the monopolization of bandwidth for ONUs with high traffic load. In order to prevent this, IPACT introduces maximum transfer window, W_max. IPACT studies on several bandwidth allocation schemes namely fixed, limited, gated, constant credit, linear credit, and elastic. Among all the six disciplines, limited scheduling discipline exhibits the best performance in IPACT. The advantages of the IPACT algorithm are it improves bandwidth utilization by reducing the overhead occurance from propagation delay and it allows statistical multiplexing. It also deploys an efficient in-

169


band signaling approach that prevents the usage of extra Ethernet frames for control. Other algorithms that do not support QoS are [9-11]. In [9], a multiple-access control scheme is proposed to provide bandwidth guaranteed (BG) service for high demand ONUs, while providing best effort (BE) service to low-demand ONUs according to the service level agreement (SLA). SLA is a contract between a network service provider and a customer that specifies what services the network service provider will furnish in measurable terms. In [10], ONUs are partitioned into two subgroups with some overlap. While frames are collected from the ONUs in one subgroup, the OLT performs DBA for ONUs in the other. Hence, the OLT continuously receives frames from the ONUs without any interruptions. In [11], REPORT messages are arranged by the request length at the next transmission cycle as long as at least one ONU requests a long enough W_max. Alternatively, when no grant length is long enough, then some requests are laid out together in the idle period to utilize the wasted idle time. The disadvantage of the DBA algorithms that do not support QoS is that they do not support the service differentiation needs of the subscribers. This causes delay in real time traffic and thus, causing the overall performance of the system to degrade. In order to support QoS, Kramer et al. propose two different methods; which are one LLID per ONU and one LLID per queue [12]. The first method allocates one LLID to the entire ONU. It presents a hierarchical scheduling structure where an OLT assigns bandwidth to the ONUs and the ONUs then further subdivide the bandwidth to the multiple queues inside the ONU. The second method allocates a single LLID to each queue. It is considered as the simplest and most robust solution where it eliminates any needs for the low-level scheduler. The scheduling will only be done in OLT, where it receives a separate REPORT message from each queue and then issue a separate GATE message for each queue.

4.2 DBA algorithms with QoS support locally Using one LLID per ONU method, Luo et al. [13] propose DBA with Multiple Services algorithm that combines limited scheduling in inter ONU allocation with non-strict priority scheduling in intra ONU allocation. Non-strict priority scheduling transmits only the previously reported packets, even if more higher-priority packets arrive after the last REPORT was sent. However, the shortcoming of non-strict

priority scheduling is increased queuing delay because all packets would have to wait a full cycle between report and transmission. In order to avoid the mentioned problem, Nikolova et al. combine limited scheduling with strict priority scheduling in [14]. In this algorithm, two types of scheduling are proposed in intra ONU allocation which are Full Priority Scheduling (FPS) and Interval Priority Scheduling (IPS). With FPS, packets are sent according to the standard strict priority scheduling. This may cause a light load punishment. Thus, IPS is proposed where it eliminates the light load punishment by using the two stage buffer explained by Kramer et al. [15]. The two stage buffers work in a way where stage I consists of multiple-priority queues and stage II consists of one FIFO queue. When a time slot arrives, packets from stage II are transmitted to the OLT, thus vacating the queue. It causes packets from stage I to be advanced into vacant spaces in the stage II queue. If the total size of stage II buffer is set to W_max, this method will ensure that the bandwidth is better utilized. But the downside of two stage buffers is that the delay may increase unboundedly. In his paper, Kramer et al. [15] also proposes another method to reduce the delay in intra ONU allocation which is to predict the high priority bandwidth requested. Prediction method has also been widely used in the literature [16–22]. In [16], Limited Sharing with Traffic Prediction algorithm has been proposed where it combines limited scheduling with traffic prediction to predict the traffic arrived during the waiting time. In [17], Early DBA mechanism is proposed where it is incorporated with an excessive bandwidth allocation scheme. Here, the prediction is done for different traffic classes and it includes the unstable degree list. In [18], IPACT with Grant Estimation (IPACT-GE) is proposed where it estimates the amount of new packets arriving between two consecutive pollings. IPACT-GE is combined with the strict priority scheduling mechanism to provide DiffServ. In [19], a model for determining the traffic burstiness-dependent optimum prediction order is proposed by using the knowledge of traffic patterns of different service classes. In [20], Local-TrafficPrediction-Based DBA algorithm is proposed in a remote-repeater-based EPON system with active forwarding. Whereas in [21], a generic QoS-aware Interleaved DBA is proposed where it combines the prediction method with the excessive bandwidth allocation to eliminate the idle period. In [22], prediction is used only for EF traffic since the nature of 170


the traffic is more constant. However the prediction method [16–22] is not due to the burstiness of the traffic. The burstiness of the traffic also caused some ONUs to have less traffic to transmit (underloaded) and some ONUs have more traffic to transmit (overloaded). For this reason, those underloaded ONUs may result in excessive bandwidth that will not be fully utilized. In order to make full use of the bandwidth, algorithms in [23-27] have been proposed to allocate the excessive bandwidth from underloaded ONUs to the overloaded ONUs. In [23] excessive bandwidth scheduling is combined with the Modified Start Time Fair Queuing (M-SFQ) technique. This technique assigns a weighting to each queue and tracks aggregate ONU service via a global virtual time. The excessive bandwidth mechanism has also been combined with two stage scheduling to reduce the delay in high priority traffic. Among them is using Modified Token Bucket (M-TB) algorithm [24]. Two stages of allocation exist in the intra ONU allocation of this algorithm, the first stage is to assign each queue according to the size of the token. Second stage is to assign the remaining bandwidth using strict priority scheduling. M-TB has also been used in [25] where a hierarchical intra ONU allocation is proposed. In [25], M-TB is combined with M-SFQ to support the QoS better. Two stage queueing has also been proposed in [26] and [27]. Penetrated EPON in [27] introduces two methods of allocation for the sub-OLT, which are EPON Dynamic Scheduling Algorithm (E-DSA) 1 and E-DSA2. E-DSA1 is used to apply excessive bandwidth scheduling to the entire ONU in both OLT allocation and sub OLT allocation, whereas E-DSA2 uses prediction method. Other than that, the excessive bandwidth mechanism is also combined with intra ONU scheduling, making it hierarchical. This can be found in [28-31]. In [28], intra ONU scheduling is done by categorizing high priority queue to normal loaded and overloaded. The high priority queue determines the grant size of each traffic class. For normal loaded high priority queue, the traffic is granted with (1:1:1) for (Expedited Forwarding (EF):Assured Forwarding (AF):BE) and for underloaded, the traffic is granted with (8:1.9:1) for (EF:AF:BE). This degrades the overall fairness of the system. In [29], the excessive bandwidth is granted according to the weight of the ONUs to reduce overgranting and to make full use of the excessive bandwidth. In [30-31] excessive bandwidth mechanism is combined with either strict priority scheduling or

non-strict priority scheduling. In [30], both strict priority and non-strict priority scheduling are proposed for the intra-ONU allocation. In [31], a non-strict priority scheduling for intra-ONU allocation is used. Excessive bandwidth combined with non-strict priority scheduling [31] is not preferred due to an increased delay to higher priority packets. Algorithms in [32] and [33] overcome this problem by proposing a strict priority scheduling in intra ONU. In Modified Smallest Available Report First (MSARF) [32], ONUs are sorted in an ascending order according to their bandwidth demands. The smallest ONUs are served first. Using excessive bandwidth mechanism in the inter ONU allocation, a minimum guaranteed bandwidth is defined for every ONU. Underloaded ONUs are granted as per request and the excessive bandwidth from these ONUs are accumulated so that it can be distributed to the overloaded ONUs according to the proportion to ensure fairness. In order to avoid overgranting, the granting of overloaded ONUs is capped up to their requested bandwidth only. For intra ONU allocation, only EF packets that have been reported in the previous REPORT message are granted first. Subsequently, AF packets are granted before BE packets including the new AF packets that arrived during EF and AF transmission. However, strict priority scheduling is known to cause light load punishment. In [33], the combination of IntServ and DiffServ is used in intra ONU. Four grants exist, where the first three grants are using IntServ, and the last grant is using DiffServ. In [34], a study of Double-Phase Polling scheduling with limited with excess distribution and shortest propagation delay first scheduling has been studied. This paper examines the implications of grant sizing and grant scheduling and the result shows that these combinations reduce the delay significantly. The excessive bandwidth mechanism is also used in intra ONU allocation rather than in the inter ONU as has been used in [35] and [36]. In Cyclic-PollingBased DBA Scheme with Service Level Agreement (CPBA-SLA) [35], inter ONU allocation is done by dividing the ONUs into three groups; Group A, B1 and B2 as according to their SLA priority. The ONUs are granted with time slots according to the ratio with two subframes exist in this algorithm. In the first subframe, while OLT schedules Group A, it collects REPORT from Group B1. Then, while OLT schedules Group B1, 171


it collects REPORT from Group A. Whereas in the second subframe, while OLT schedules Group A, it collects Group B2. Inside the ONU, bandwidth allocation is done based on the excessive bandwidth mechanism to each type of DiffServ traffic. Excessive bandwidth mechanism is also used in intra ONU allocation of [36], but with limited IPACT in inter ONU allocation. The excessive bandwidth mechanism is also used in 10G EPON [37], long reach EPON [38] and hybrid time division multiplexing (TDM)/WDM EPON [39]. The downside of excessive bandwidth method [23-39] is that it may overgrant some of the overloaded ONUs or queues and increases the idle period of overloaded ONUs or queues if the allocation of excessive bandwidth is not conducted fairly. Other algorithms that support QoS locally can be found in [40–44]. In [40], unlike IPACT, rather than having a fixed polling order, its polling order varies. In inter ONU allocation, the polling order is first determined at the beginning of the transmission cycle. The longest queue will then be polled first. For intra-ONU allocation, ONU will first send all the highest priority messages. If it has reached a threshold, it will not consider the lower priority traffic. In [41], Zhu et al. proposes three different schemes for the OLT allocation and a two stage scheduling scheme in ONU allocation. For inter ONU allocation, parameter based Call Admission Control (CAC) mechanism is used to decide which ONU belongs to BG or non-BG and determines how many entries each BG ONU is allocated. Based on the results of CAC, the Even Distribution Algorithm (EDA) generates the entry table and the non-BG ONU list to set the polling sequence. Then, the OLT polls ONUs one after another based on the results of EDA, inviting ONUs to transmit data over the channel. In [42], Dhaini et al. propose the first framework for per-stream QoS protection using a two-stage Admission Control (AC) system. The first stage enables the ONU to perform flow admission locally according to the bandwidth availability and the second stage allows for global AC. Paper in [43] proposes a mathematical model for the framework in [42]. In [44], Enhanced Burst-Polling DBA method is proposed where it specifies that every ONU is polled periodically in a burst manner. It adaptively increases or decreases the minimum guaranteed bandwidth of the

three traffic classes according to the requested bandwidth of an ONU. In [45], Yin et al. proposes User-oriented Hierarchical bandwidth Scheduling Algorithm that has two hierarchical level. For inter-ONU allocation, an improved hybrid cycle approach that separates a frame into a static part for EF traffic and an adaptive dynamic part for AF and BE traffic are proposed. Whereas the intra ONU allocation adopts credit-based scheduling approach where a credit-based common queue is combined with a transmission priority scheme to enhance scheduling architecture and minimize the average number of queues in the ONU. One of the disadvantages of supporting QoS locally is that it puts intelligence in both OLT and ONU, and thus adds the complexity of the algorithm. More importantly, the QoS can only be supported either inside the OLT or inside the ONU. Thus, it does not ensure fairness of the entire system. The overall performance can also be degraded. In order to ensure that QoS is supported in the entire system, algorithms that support QoS globally is preferred. Recently, fuzzy logic has been used in allocating the bandwidth [46] and [47]. In [46], a Fuzzy PredictionBased DBA algorithm is proposed with Fuzzy Unstable Degree List Controller (FUDLC) and Fuzzy Credit Estimator (FCE) mechanisms incorporated to improve the prediction accuracy. The FUDLC chooses the second traffic variance and the mean traffic variance of ONUs as input linguistic variables to determine the optimal number of ONUs in the unstable degree list. In addition, the FCE chooses the degree of traffic variance and the degree of waiting time among ONUs as input linguistic variables for the credit estimation, so that the request bandwidth for the next cycle can be predicted more precisely. In [47], fuzzy logic controller is proposed where it can be implemented as a simple look-up table to making EPON ideal for high-speed operation. Four different metrics are considered in this algorithm which are the delay of the head-of-the-line packets at the ONUs, the importance level of the packets, the relative ONU’s buffer fullness, and the level of the power fluctuation from one ONU to another. However, the fuzzy logic method may add up to the complexity of the design and the system, thus causing the cost and delay to be higher. The disadvantages of supporting QoS locally in general are that it may increase the complexity of the system 172


since the intelligence is placed at both OLT and ONUs and it also increases the queuing delay. Besides that, overgranting may happen when the excessive bandwidth from overloaded ONUs are not managed properly.

Table 1. Summary of DBA schemes No 1

Scheme Limited scheduling adaptive cycle time

2

Limited with non-strict priority scheduling Limited with strict priority scheduling Traffic prediction

4.3 DBA algorithms with QoS support locally Supporting QoS globally offers many advantages especially to the real time traffic. The delay and jitter of high priority traffic can be minimized and thus the overall performance is enhanced. Among the algorithms that support QoS globally can be found in [48-52]. Algorithm in [48] consists of two parts. The first part is inter-ONU scheduling, which is a delta DBA based on burst polling, and the second part is OLT based intra-ONU scheduling that is responsible for a differentiated priority queuing. The concept of this algorithm is that it divides ONUs into underloaded and overloaded ONUs. The excessive bandwidth of underloaded ONUs are allocated to every queue according to the weight of each queue. Other than that, a two cycle allocation scheme is proposed in [49]. The schemes are Grant Before REPORT that is used to allocate EF bandwidth via prediction method and Grant After REPORT that allocates AF and BE bandwidth. Class-of-service (CoS) oriented packet scheduling [50] regulates the traffic of each ONU and CoS using two sets of credit pools, one per ONU and one per CoS. Firstly, each ONU with traffic for the current CoS is granted up to the number of credits stored for that ONU-CoS. Next, the unused credits are pooled together and distributed to the CoS-ONU pairs that were not fully satisfied. In [51], a Proportional Sharing with Load Reservation algorithm provides bandwidth that guarantees per-flow basis and redistributes the unused bandwidth among active flows in proportion to their priority level. In [52] excessive bandwidth scheduling is combined with the M-SFQ technique by limiting the granted bandwidth to requested. This reduces the light load punishment in M-SFQ method. The advantage of supporting QoS globally is that it can ensure that QoS is met globally. However, the downside is that this method could have higher scheduling overhead because each queue requires separate messages. Summary of the classificationof the algorithm is shown in Table 1.

3

4

5

Excessive bandwidth

6

Excessive bandwidth with non-strict priority scheduling Excessive bandwidth with strict priority scheduling Excessive bandwidth in intra ONU allocation Fuzzy logic

7

8

9 10

Support QoS globally

Characteristics Improves bandwidth utilization by reducing the overhead occurrence from propagation delay and it allows statistical multiplexing but may not support multi service traffic. Support QoS but may cause delay in high priority traffic.

Support QoS but may cause light load punishment Shorter delay but prediction may be less accurate if not properly structured due to the burstiness of the traffic Improves bandwidth utilization but may not support multi service traffic. Support QoS but may cause delay in high priority traffic.

Support QoS but may cause light load punishment

Shorter delay but support QoS only locally.

Improves bandwidth utilization but complex Support QoS globally but if not structured properly may cause overgranting

5 EPON TESTBED The first testbed study of EPON that can be found in the literature is [53] where it proposes a clock automatic protection switching (APS) scheme for resolving the clock problem in the EPON. This testbed study is done by using field programmable gate array (FPGA). The studies on the synchronization of EPON are done in [54] where it uses EP2C70F896 FPGA chip to implement the IEEE 1588 version 2 protocol. The FPGA has also been used by other researchers in EPON testbed [55–58]. The FPGA is used in [55] for

173


the OLT design. It uses a Cyclone EP1C20F324C6 FPGA chip for the OLT. On the other hand, the Stratixll EP2S180 FPGA chip is used in [56] to transmit forward error correction (FEC)-coded frames in physical coding sublayer for Ethernet PON. Cyclone III EP3C25Q240C8 FPGA chip is used in designing the ONU in [57] where it allows the performing of auto-discovery mode in EPON. A low cost ONU has been designed in [58] that uses a new generation of low-cost XILINX FPGA based on Spartan 6. Survivable EPON system is proposed in [59] using two trees and an original centralized APS control mechanism implemented in FPGA. Paper in [60] is proposed to introduce the basic framework of EPON by using Teknovus chips in the design. TK3721 is used as the OLT kernel chip and TK3713 is used as the ONU kernel chip. This testbed is based on embedded Vxworks platform. Triple play over EPON testbed has been established in Taiwan for EPON function and performance testing. Together with EPON field trial, the research was completed in [61]. Both OLT and ONUs are developed by Industrial Technology Research Institute (ITRI). Ethernet switch has also been used as OLT and ONU in the EPON testbed proposed in [62]. The testbed includes EPON access network and metro core network. The first implementation of the DBA in EPON testbed that can be found in the literature is in [63]. This paper proposes a Request Counter-DBA (RC-DBA) algorithm that concentrates the allocation of the bandwidth at the OLT. OLT is designed using Cyclone EP1C12F324C8 FPGA where the RC-DBA algorithm is implemented in it. A hierarchical DBA is proposed in Two Threshold reporting and Strict Priority (TTSP) [64] where limited scheduling is used in inter ONU allocation and strict priority scheduling is used in intra ONU allocation. This hierarchical DBA is also implemented in FPGA. Hierarchical DBA is also proposed in the HierarchicalWeighted Round Robin (H-WRR) [65] where a scheduler cyclically visits the entries belonging to every class and allocates grant lengths according to the received queue reports and the token sizes. This algorithm is implemented inside ASIC chip.

6 CONCLUSION Various studies have been done on the DBA as it becomes one of the crucial issues in EPON. In this

paper, we discuss on EPON and its protocol as well as the classification of the DBA algorithms. A significant review proposed by several authors has been listed out throughout this paper for all DBA methods. We also highlight theirs advantages and disadvantages for the references of future researches. It can be concluded that very classification has their pros and cons and researches may choose the classification that is suitable to their needs.

7 REFERENCES 1. Yang, K., Ou, S., Guild, K., Chen, H.-H.: Convergence of Ethernet PON and IEEE 802.16 broadband access networks and its QoS-aware dynamic bandwidth allocation scheme, IEEE Journal on Selected Areas in Communications, vol. 27, no. 2, pp. 101 –116, (2009). 2. IEEE 802.3ah task force home page, IEEE P802.3ah Ethernet in the first mile task force. [Online]. Available: http://www.ieee802.org/3/efm, (2004). 3. Xiaofei, J., Ge, F.: Performance analysis of a halfduplex EPON system, ACTA Photonica Sinica, vol. 32, p. 6, (2003). 4. Desai, B., Frigo, N., Smiljanic, A., Reichmann, K., Iannone, P., Roman, R.: An optical implementation of a packet-based (Ethernet) MAC in a WDM passive optical network overlay, in Optical Fiber Communication Conference, Optical Society of America (2001). 5. Zhang, Z., Xu, J., Liu, D., Zhu, G.: WDM-based gigabit Ethernet PON system, in Proc. SPIE 5282, vol. 210, (2004). 6. Kramer, G., Pesavento, G.: Ethernet passive optical network (EPON): Building a next-generation optical access network, IEEE Communications Magazine, vol. 40, no. 2, pp. 66–73, (2002). 7. Kramer, G., Mukherjee, B., Pesavento, G.: Ethernet PON (PON): Design and analysis of an optical access network, Photonics Networking Communications, vol. 3, pp. 307–319, (2001). 8. Kramer, G., Mukherjee, B., Pesavento, G.: IPACT a dynamic protocol for an Ethernet PON (EPON), IEEE Communications Magazine, vol. 40, pp. 74–80, (2002). 9. Ma, M., Zhu, Y., Cheng, T.: A systematic scheme for multiple access in Ethernet passive optical access networks, Journal of Lightwave Technology, vol. 23, no. 11, p. 3671, (2005). 10. Choi, S., Lee, S., Lee, T., Chung, M., Choo, H.: Doublephase polling algorithm based on partitioned ONU subgroups for high utilization in EPONs, IEEE/OSA Journal of Optical Communications and Networking, vol. 1, no. 5, pp. 484–497, (2009). 11. Chen, W., Wang, W., Hwang, W.: Adaptive dynamic bandwidth allocation algorithm with sorting report messages for Ethernet passive optical network, IET Communications, vol. 4, no. 18, pp. 2230–2239, (2010). 12. Kramer, G.: On configuring logical links in EPON, Teknovus, Inc, (2001).

174

International Journal of New Computer Architectures and their Applications (IJNCAA) 4(4): 167-176 The Society of Digital Information and Wireless Communications, 2014 (ISSN: 2220-9085) 13. Luo, Y., Ansari, N.: Bandwidth allocation for multiservice access on EPONs, IEEE Communications Magazine, vol. 43, no. 2, pp. S16–S21, (2005). 14. Nikolova, D., Houdt, B. V.; Blondia, C.: Dynamic bandwidth allocation algorithms for Ethernet Passive Optical Networks with threshold reporting, Springer Telecommunication Systems, vol. 28, pp. 31–52, (2005). 15. Kramer, G., Mukherjee, B., Dixit, S., Ye, Y., Hirth, R.: Supporting differentiated classes of service in Ethernet passive optical networks, Journal of Optical Networking, vol. 1, pp. 280–298, (2002). 16. Luo Y., Ansari, N.: Limited sharing with traffic prediction for dynamic bandwidth allocation and QoS provisioning over Ethernet passive optical networks, Journal of Optical Networking, vol. 4, no. 9, pp. 561–572, (2005). 17. Hwang, I.-S., Shyu, Z.-D., Ke, L.-Y., Chang, C.-C.: A novel Early DBA mechanism with prediction-based fair excessive bandwidth allocation scheme in EPON, Computer Communications, vol. 31, no. 31, pp. 1814– 1823, (2008). 18. Zhu, Y., Ma, M.: IPACT with Grant Estimation (IPACT-GE) scheme for Ethernet passive optical networks, Journal of Lightwave Technology, vol. 26, no. 14, pp. 2055– 2063, (2008). 19. De, S., Singh, V., Gupta, H., Saxena, N., Roy, A.: A new predictive dynamic priority scheduling in Ethernet passive optical networks (EPONs), Optical Switching and Networking, vol. 7, no. 4, pp. 215–223, (2010). 20. Aun Chan, C., Attygalle, M., Nirmalathas, A.: Localtraffic redirection-based dynamic bandwidth assignment scheme for EPON with active forwarding remote repeater node, IEEE/OSA Journal of Optical Communications and Networking, vol. 3, no. 3, pp. 245–253, (2011). 21. Hwang, I., Lee, J., Lai, K., Liem A.: Generic QoSAware interleaved dynamic bandwidth allocation in scalable EPONs, IEEE/OSA Journal of Optical Communications and Networking, vol. 4, no. 2, pp. 99–107, (2012). 22. Li, X., Dan, L., Wu, Q.: Adaptive dynamic bandwidth allocation algorithm supporting multi-services over Ethernet passive optical networks, Optik-International Journal for Light and Electron Optics, (2012). 23. Ghani, N., Shami, A., Assi, C., Raja M.: Intra-ONU bandwidth scheduling in Ethernet passive optical networks, IEEE Communications Letters, vol. 8, no. 11, pp. 683–685, (2004). 24. Chen, J., Chen, B., He, S.: A novel algorithm for intraONU bandwidth allocation in Ethernet passive optical networks, IEEE Communications Letters, vol. 9, no. 9, pp. 850–852, (2005). 25. Chen, B., Chen, J., He, S.: Efficient and fine scheduling algorithm for bandwidth allocation in Ethernet passive optical networks, IEEE Journal of Selected Topics in Quantum Electronics, vol. 12, no. 4, pp. 653–660, (2006). 26. Li, Q., Li, H., Ji, Y.: Flexible scheduling ensuring interONU-group fairness in 1G/10G EPON coexistence system, The Journal of China Universities of Posts and Telecommunications, vol. 16, no. 4, pp. 16–22, (2009). 27. Shami, A., Bai, X., Ghani, N., Assi, C., Mouftah H.: QoS control schemes for two-stage Ethernet passive optical

access networks, IEEE Journal on Selected Areas in Communications, vol. 23, no. 8, pp. 1467–1478, (2005). 28. Yang, Y., Nho, J., Mahalik, N., Kim, K., Ahn, B.: A traffic-class burstpolling based delta DBA scheme for QoS in distributed EPONs, Computer Standards & Interfaces, vol. 28, no. 6, pp. 721–736, (2006). 29. Bai, X., Shami, A., Assi, C.: On the fairness of dynamic bandwidth allocation schemes in Ethernet passive optical networks, Computer Communications, vol. 29, no. 29, pp. 2123–2135, (2006). 30. Lee, H., Yoon, W., Lee, T., Choo, H., Chung, M.: TwoPhase Cycle DBA (TCDBA) for Differentiated Services on EPON, IEICE transactions on communications, vol. 92, no. 9, pp. 2823–2837, (2009). 31. Sue, C., Sung, W.: Fitting Scheduling Timing–Elastic Weighted Granting (FST–EWG): An EPON DBA Algorithm, Journal of Optical Communications and Networking, vol. 4, no. 6, pp. 468–479, (2012). 32. Dolama, M., Rahbar, A.; Modified Smallest Available Report First: New dynamic bandwidth allocation schemes in QoS-capable EPONs, Optical Fiber Technology, vol. 17, no. 1, pp. 7–16, (2011). 33. Liu, H., Li, Y., Ji, J., Qiao, Y.: Resources allocation in an Intserv/Diffserv integrated EPON system, The Journal of China Universities of Posts and Telecommunications, vol. 16, no. 3, pp. 108–121, (2009). 34. McGarry, M., Reisslein, M.: Investigation of the DBA algorithm design space for EPONs, Journal of Lightwave Technology, vol. 30, p. 14, (2012). 35. Choi, S., Park, J.: SLA-aware dynamic bandwidth allocation for QoS in EPONs, Journal of Optical Communications and Networking, vol. 2, no. 9, pp. 773– 781, (2010). 36. Merayo, N., Dur´an, R., Lorenzo, P. F. R., Ignacio, M., Abril, E.: EPON bandwidth allocation algorithm based on automatic weight adaptation to provide client and service differentiation, Photonic Network Communications, vol. 17, pp. 119–128, (2009). 37. Lim, W., Yang, Y., Milosavljevic, M.: Multicast polling for 10G-EPON, Electronics Letters, vol. 48, no. 9, p. 513, (2012). 38. Kantarci, B., Mouftah, H.: Delay-constrained admission and bandwidth allocation for long-reach EPON, Journal of Networks, vol. 7, p. 5, (2012). 39. Dhaini, A., Assi, C., Maier, M., Shami, A.: Dynamic wavelength and bandwidth allocation in hybrid TDM/WDM EPON networks, Journal of Lightwave Technology, vol. 25, no. 1, pp. 277–286, (2007). 40. Maode Ma, L. L., Cheng, T. H.: Adaptive scheduling for differentiated services in an Ethernet passive optical network, Journal of Optical Networking, vol. 4, pp. 661– 670, (2005). 41. Zhu, Y., Ma, M., Cheng, T.: Hierarchical scheduling to support differentiated services in Ethernet passive optical networks, Computer Networks, vol. 50, no. 3, pp. 350–366, (2006). 42. Dhaini, A., Assi, C., Maier, M., Shami, A.: Per-stream QoS and admission control in Ethernet passive optical

175

International Journal of New Computer Architectures and their Applications (IJNCAA) 4(4): 167-176 The Society of Digital Information and Wireless Communications, 2014 (ISSN: 2220-9085) networks (EPONs), Journal of Lightwave Technology, vol. 25, no. 7, pp. 1659–1669, (2007). 43. Assi, C., Maier, M., Shami, A.: Toward quality of service protection in Ethernet passive optical networks: challenges and solutions, IEEE Network, vol. 21, no. 5, pp. 12–19, (2007). 44. Lim, W., Yun, C., Yang, Y., Lee, H., Kim, K.: Burstpolling based dynamic bandwidth allocation using adaptive minimum guaranteed bandwidth for EPONs, IEEE/OSA Journal of Optical Communications and Networking, vol. 1, no. 7, pp. 594–599, (2009). 45. Yin Y., Poo, G.: User-oriented hierarchical bandwidth scheduling for Ethernet passive optical networks, Computer Communications, vol. 33, no. 8, pp. 965–975, (2010). 46. Hwang, I.-S., Shyu, Z.-D., Huang, K.-K.: Fuzzy logic embedded in prediction-based DBA for differentiated services on EPONs, Tamkang Journal of Science and Engineering, vol. 12, no. 1, pp. 73–84, (2009). 47. Razavi, R., Guild, K.: Multiconstraints fuzzy-logicbased scheduling algorithm for passive optical networks, Journal of Optical Networking, vol. 8, pp. 346–357, (2009). 48. Yang, Y., Ahn, B., Nho, J.: Supporting quality of service by using delta dynamic bandwidth allocations in Ethernet passive optical networks, Journal of Optical Networking, vol. 4, no. 2, pp. 68–81, (2005). 49. Shami, A., Bai, X., Assi, C., Ghani, N.: Jitter performance in Ethernet passive optical networks, Journal of Lightwave technology, vol. 23, no. 4, p. 1745, (2005). 50. Naser, H., Mouftah, H.: A joint-ONU interval-based dynamic scheduling algorithm for Ethernet passive optical networks, Networking, vol. 14, no. 4, pp. 889–899, (2006). 51. Pereira, F., da Fonseca, N., Arantes, D.: A fair scheduling discipline for Ethernet passive optical networks, Computer Networks, vol. 53, no. 11, pp. 1859–1878, (2009). 52. Chen, J., Chen, B., Wosinska, L.: Joint bandwidth scheduling to support differentiated services and multiple service providers in 1G and 10G EPONs, Journal of Optical Communications and Networking, vol. 1, no. 4, pp. 343– 351, (2009). 53. Yang, L., Jiao, M.: Synchronization solution with clock automatic protection switching ability for the Ethernet passive optical network, Journal of Optical Networking, vol. 3, no. 10, pp. 736–741, (2004). 54. Wang, Z., Wang, M., He, K., Chen, X.: Implementation of EPON systems supporting accurate time synchronization, in Third International Conference on Communications and Mobile Computing (CMC), pp. 210–212, (2011). 55. Zou, J., Lin, R., Liu, M.: Design and FPGA implementation of OLT for EPON, in 6th International Conference on ASIC, vol. 2, pp. 761–764, (2005). 56. Yen, C.: An efficient PCS to transmit FEC-coded frames in Ethernet PON, in International Symposium on High Capacity Optical Networks and Enabling Technologies, pp. 245–250, (2008). 57. Mady, A., Tonini, A.: A VHDL implementation of ONU auto-discovery process for EPON, in International Conference on Networking and Media Convergence, pp. 36– 41, (2009).

58. Zhang, F., Liu, Z., Zhao, Y., Zhang, Y., Ma, C.: EPON ONU system design and implementation based on Spartan 6, in International Conference on Computer Science and Service System (CSSS), pp. 2226–2229, (2011). 59. Wang, Z., Wang, L., Chen, X.: Implementation of a survivable EPON system supporting direct communication between ONUs,” in 19th Annual Wireless and Optical Communications Conference (WOCC), pp. 1–4, (2010). 60. Jie, S., Junni, Z., Jiajun, Y., Xuping, Z.: Design and implement of practical EPON network management system, in Asian Optical Fiber Communication Optoelectronic Exposition Conference, pp. 1 –3, (2006). 61. Lin, T., Fan, K.: EPON testbed and field trial environment in Taiwan, in 2nd International Conference on Testbeds and Research Infrastructures for the Development of Networks and Communities, pp. 7–pp, (2006). 62. Liu, Y., Phung, M., Chua, K., Mohan, G., Guo, Y., Panneerselvan, S.: A unified network control and management system for an integrated EPON and metro optical network testbed, in 3rd International Conference on Testbeds and Research Infrastructure for the Development of Networks and Communities, pp. 1–6, (2007). 63. Jang, J., Kang, H., Jeon, H.: The FPGA implementation of the RCDBA algorithm in the EPON network, International Journal of Computer Science and Network Security, vol. 6, no. 6, pp. 44–51, (2006). 64. Bin, L., Chen, X., Gao, R., Lu, X.: A dynamic bandwidth allocation scheme with two threshold reporting and strict priority scheduling for multiservice EPONs, in First International Conference on Communications and Networking in China, pp. 1–5, (2006). 65. Kim, C., Yoo, T., Kim, B.: A hierarchical weighted round robin EPON DBA scheme and its comparison with cyclic water-filling algorithm, in IEEE International Conference on Communications, pp. 2156–2161, (2007).

176