Sie sind auf Seite 1von 12

IEEE/OSA JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. XX, NO.

XX, NOVEMBER 2004 1

Dynamic Bandwidth Allocation Schemes


to Improve Utilization under Non-Uniform Traffic
in Ethernet Passive Optical Networks
Kyuho Son, Hyungkeun Ryu, Student Member, IEEE and Song Chong, Member, IEEE

Abstract— Ethernet Passive Optical Networks (EPONs) are an must share a single uplink optical fiber trunk. To prevent data
emerging access network technology that provides a low cost from being corrupted due to multiple ONUs transmitting at
solution for fiber-to-the-home (FTTH) and fiber-to-the-business the same time, a medium access control protocol is required.
(FTTB). In this paper, conventional bandwidth allocation schemes
in EPON are shown to suffer from poor utilization under non- Therefore, a Multi-Point Control Protocol (MPCP) is being
uniform traffic conditions, particularly as the number of ONUs, developed by the IEEE 802.3ah task force to arbitrate between
guard time and round-trip time increase. To resolve this problem, transmissions by ONUs to avoid collisions [4]. The MPCP
we propose a new scheme that allocates a timeslot intelligently is operated principally by two control messages, GATE and
by considering other ONUs’ queue occupancy, instead of strictly REPORT. Each ONU informs the OLT of the queue occupancy
enforcing a maximum timeslot size. The analysis and simulation
results show that the proposed scheme can provide significantly by REPORT to help the OLT make a decision about bandwidth
higher utilization than conventional schemes and can support allocation. The OLT assigns the transmission timeslot by
max-min fairness under non-uniform traffic conditions. GATE.
Index Terms— access network, Ethernet Passive Optical Net- In EPON, one of the most important issues is how to share
work (EPON), bandwidth allocation scheme, non-uniform traffic, uplink resources efficiently and fairly (i.e., the bandwidth
maximum utilization, max-min fairness, performance analysis. allocation problem), and much research is currently being
done by experts on this topic. Kramer et al. [5] suggested
a fixed service scheme that always grants the maximum
I. I NTRODUCTION
timeslot size to each ONU. They determine the optimal max-
Recently, the capacity of backbone networks has been imum timeslot size by using a utilization-delay optimization
increased dramatically. However, in access networks, dial-up, approach. In Interleaved Polling with Adaptive Cycle Time
xDSL(x Digital Subscriber Line) and cable modem technolo- (IPACT) [6], they also proposed a well-designed bandwidth
gies provide relatively small increases in data transfer. With the allocation framework that minimizes the unused timeslot by
rapid growth of the number of Internet users and multimedia interleaving polling messages. IPACT can reduce the polling
services, access networks have become bottlenecks. A passive delay while maintaining high link utilization, and can be used
optical network (PON), which is a point-to-multipoint optical with a variety of allocation schemes: fixed, gated, limited,
network composed solely of passive elements, is regarded constant/linear credit service, etc. Kramer et al. compared the
as the technology most likely to offer a solution to this above schemes through simulations and concluded that neither
problem, [1], [2] and is considered to be the best way to sup- of the discussed service schemes is better than the limited
port emerging services such as high-quality digital television service scheme. In [7], the authors introduced the concept
broadcast services and online education multicast services. In of threshold report to achieve higher bandwidth efficiency.
particular, Ethernet PON (EPON) is the best candidate for In addition to research on the bandwidth allocation problem,
a next generation optical access network because Ethernet is there is much research being done on differentiating quality
cheap, simple, scalable and popular, and is regarded as an of services (or classes of services) in EPON [8]–[11].
economical method for deploying fiber-to-the-X (FTTX) [3]. Previous research mostly has performed simulations under
Fig. 1 shows a typical topology of EPON. An optical line the assumption of uniform traffic conditions. In this paper,
terminal (OLT), located at a local exchange, is connected we show that not only do conventional bandwidth allocation
to multiple optical network units (ONUs) through passive schemes suffer from utilization degradation under non-uniform
elements, splitter and coupler. In downlink transmission from traffic conditions, but also that the degradation can be se-
OLT to ONUs (using 1550nm wavelength), a packet is broad- vere when the number of ONUs, guard time and round-trip
casted automatically to all ONUs through an optical splitter time increase. We note that this problem can be resolved if
and each ONU filters the received packet according to its an intelligent decision is made about bandwidth allocation
destination address. In uplink transmission from ONUs to without maximum timeslot restriction, which decision takes
OLT (using 1310nm wavelength), a packet from all ONUs into account not just one, but all queues of ONUs. Our pro-
can only reach the OLT, not other ONUs. As a result, ONUs posed scheme can provide higher utilization than conventional
schemes and can support max-min fairness under non-uniform
Manuscript received January xx, 2004; revised November xx, 2004. This
work was supported in part by University IT Research Center Project and traffic conditions.
Electronics and Telecommunications Research Institute, Korea. The remainder of this paper is organized as follows. In
IEEE/OSA JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. XX, NO. XX, NOVEMBER 2004 2

TABLE I
III. BANDWIDTH A LLOCATION S CHEMES
D ETAILED DESCRIPTION OF GATE SCHEDULING .
In this section, we describe conventional and proposed
Every time the OLT receives REPORT from ONUi , bandwidth allocation schemes. We consider the following to
1) Update SEI ← max(SEI, now + RT Ti ). be basic bandwidth allocation schemes.
start time = SEI 1) Fixed service scheme always grants the maximum times-
2) Decide the timeslot size Gi by the specific bandwidth lot size Gmax .
allocation scheme. 2) Gated service scheme grants as much as is requested,
duration = Gi Gk+1
i = Rik .
3) Update SEI ← SEI + Gi + r + g.
3) Limited service scheme grants as much as is requested,
4) Send GATE to ONUi . but cannot exceed Gmax , Gk+1
i = min(Rik , Gmax ).
where Gki and Rik denote the kth timeslot size for ONUi and
the kth request (i.e. queue length information) from ONUi
section 2, we describe an EPON system model. In section 3, respectively.
we present conventional and proposed bandwidth allocation Various bandwidth allocation schemes, including the three
schemes. In sections 4 and 5, we state the utilization analysis schemes stated above, are compared in [6]. The fixed service
and various simulation results, to demonstrate that our schemes scheme is simple, but does not take into account the amount
perform well. We present conclusions in section 6. of traffic from each ONU. This scheme is mentioned here
only for comparison. The authors find that the gated service
scheme can utilize the bandwidth more efficiently without any
II. M ODEL D ESCRIPTION unused remainder4 , and the delay and average queue length
of ONUs are much smaller than in other schemes in the mid-
Assume that an EPON has BW link capacity and consists range offered load. However, the gated service scheme is not
of an OLT and ONUi , 1 ≤ i ≤ N , where N is the total number suitable for use in EPON because the cycle time and delay
of ONUs in the EPON. Every time the OLT receives REPORT may increase without bound as the load increases. Therefore,
containing a request (i.e., queue length) from ONUi , it sends the authors conclude that neither of the schemes discussed in
GATE containing the transmission start time and duration [6] is better than the limited service scheme. In this paper, we
to ONUi . A detailed description and graphical explanation will compare our proposed schemes with the best-performance
of this GATE scheduling1 are given in Table I and Figure 2, limited service scheme.
respectively. SEI (Scheduling End-point Indicator) is the last The limited service scheme can achieve higher utilization,
point that has been allocated (i.e., the earliest point that a new but may fail to fully utilize the bandwidth under non-uniform
allocation can be started) and now is the current time. RT Ti traffic conditions. Suppose that only one ONU wants to send
is the round-trip time2 between the ONUi and the OLT. r is data. The ONU can use Gmax in each cycle, where a cycle
the time required to send REPORT3 , and g is the guard time consists of Gmax and a number N of guard times and report
including the following: the ONU’s laser turn on/off time, and times. Therefore the utilization is limited to GmaxG+N (r+g) .
max

the OLT’s receiver auto gain control (AGC) settling time and Assuming that the number of ONUs N = 16, maximum
clock recovery time. timeslot size Gmax = 125µs, guard time g = 5µs and report
When ONUi receives GATE from the OLT, it sends data time r = 0.512µs, the maximum utilization is only 58.6%.
throughout the allotted time Gi from start time, and then The situation becomes more serious as the number of ONU
sends a new REPORT, which contains current queue length N and the guard time g increase. This is because the OLT
information Ri , at the end of data transmission. If the OLT cannot grant more than Gmax even if other ONUs do not
receives a new REPORT from ONUi , it performs the GATE use any resources. We can expect utilization improvements if
scheduling repeatedly. the bandwidth allocation scheme can grant a timeslot longer
In the ordinary case of SEI ≥ now +RT Ti in Figure 2 (a), than Gmax considering other ONUs’ traffic. Therefore, two
the OLT allocates all the timeslots without omission. However, proposed schemes that satisfy this condition, P1 and P2, are
in the case of SEI < now + RT Ti in Figure 2 (b), the OLT suggested in this paper.
cannot receive data from ONUi directly after the current SEI
even if the OLT sends GATE immediately, because the OLT
must wait for a previous REPORT to send a new GATE. Due A. Proposed Scheme 1 (P1)
to this unallocated timeslot, utilization degradation may occur. Using other ONUs’ GATE information Gj for i 6= j (the
In addition, the cycle time and the delay cannot be smaller than recently granted timeslots of other ONUs), P1 calculates how
a certain value even if the load is extremely small. We refer many timeslots have already been used by other ONUs and
to this phenomenon as the RTT effect. how many timeslots remain. To perform this task, we newly
define a variable Fik and set its value to the total timeslots
1 Our GATE scheduling method is quite similar to the method proposed in
IPACT [6]. 4 In the fixed and limited service schemes, an unused remainder of the
2 In EPON, the geographical distance from an OLT to ONUs can be from
granted timeslot may exist at the end of transmission if an ONU has more
several km to 20km [4] and the RTT can be from several µs to 200 µs. packets to send than the granted timeslot. This is because the packet size at
3 Fixed service scheme do not need REPORT message, r = 0. the head of the queue may be bigger than the remainder.
IEEE/OSA JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. XX, NO. XX, NOVEMBER 2004 3

per cycle minus the timeslots recently used by other ONUs. where Ri is the recent request from ONUi and si indicates
This variable indirectly indicates how much OLT can grant to whether or not the OLT grants as much as ONUi requests.
ONUi . Suppose that only one ONUi always has enough data to
When Rik ≤ Gmax , P1 grants as much time as requested, transmit and other ONUs do not have any data. This means
k
Ri , as in the limited service scheme. Otherwise, P1 guarantees that Rj = 0, sj = 1, j 6= i and si = 0, thus OLT can grant
N Gmax
+N (r+g) =
at least Gmax and attempts to grant Ri as much as possible N Gmax to ONUi and the utilization will be N Gmax
but no more than Fik . As a result, the utilization performance 95.8%.
of P1 is at least as good as that of the limited service scheme. Some features of the four bandwidth allocation schemes
 k
Ri if Rik ≤ Gmax (fixed, limited, P1 and P2) under the non-uniform traffic are
k+1
Gi = k k summarized in Table. II.
min(Ri , max(Gmax , Fi )) otherwise
(1)
P(i−1) mod N
where Fik = N Gmax − j=(i−N +1) mod N Gj . IV. P ERFORMANCE A NALYSIS
where Gi is the size of the timeslot recently allotted to ONUi . In this section, we analyze the utilization performance of
The utilization will be improved by relaxing the constraint on the fixed, limited and proposed schemes. First, we will derive
the maximum size of timeslot Gmax . For example, when only the utilization under arbitrary ONUs’ input traffic, assuming
one ONUi wants to send data, the OLT can grant N Gmax to that there is no RTT effect (this does not mean that RT T = 0).
N Gmax Next, we will find the maximum utilization under the special
ONUi and the utilization will be N Gmax +N (r+g) = 95.8%.
But P1 may cause a problem regarding the fairness of times- non-uniform scenario in which the traffic load of ONU1 is
lot allocation, because it is possible that an ONU, which was 1 − ρ and the traffic load of ONU2∼N is ρ. Finally, we will
originally using the N Gmax timeslot, yields not N G2max , but take the RTT effect into account.
only Gmax to a newly arriving ONU. For better presentation,
we depict this example in Fig. 3 and compare P1 with other
A. Utilization without considering the RTT effect
schemes.
Let Ii and Oi respectively denote the rate of input and
output at ONUi normalized to EPON link rate. We will derive
B. Proposed Scheme 2 (P2) output rates {Oi }i=1,2,...,N in terms of arbitrary input rates
We now propose a new dynamic bandwidth allocation {Ii }i=1,2,...,N without considering the RTT effect. Then, the
scheme (P2) to achieve both utilization improvement and max- PN
total link utilization will be U = k=1 Oi .
min fairness. P1 uses ONUs’ GATE information, while P2 uses 1) Fixed service scheme: The OLT always grants Gmax .
ONUs’ REPORT information (the recently requested queue Thus, the cycle time is equal to a constant c = N Gmax + N g.
length of other ONUs). The output rate can be calculated as
To calculate fair distribution, if an OLT sends GATEs  
to all ONUs once the OLT has received requests from all Gmax
Oi = min Ii , (3)
ONUs, the polling time will cause timeslot waste. Therefore, a N Gmax + N g
scheme in which the OLT does not wait until all the requests
and the output rate is the same as the input rate for all ONUi
are collected is preferable. In the time between the OLT’s
(i.e. the system is stable) when
receiving REPORT from ONUi and sending the corresponding
GATE, the OLT will receive REPORT from a number M Gmax
Ii ≤ , f or all i. (4)
of other ONUs, and will not send GATE to them before N Gmax + N g
it sends GATE to ONUi . Hence, the REPORT information
2) Limited service scheme: To determine whether or not
from M ONUs is not used. However, the received REPORT
si = 1, assume that si = 1. As we mentioned earlier, si
information from the other ONUs (total number of ONUs N
denotes whether or not the OLT grants as much as ONUi
minus M ) is used, in other words, the OLT sends GATE to
requests. If we replace input rate Ik , which is larger than Ii ,
(N − M ) ONUs before it sends GATE to ONUi . In P2, the
by Ii or Ik0 = min(Ii , Ik ), then all ONUs will satisfy sk = 1,
OLT distributes N Gmax max-min fairly among ONUs based
Ok0 = Ik0 for all k. Under the changed input rates, the cycle
on total N requests and sends GATE to ONUi with the amount (r+g)
time will be c0 = NP N because the total input rate is
of its own fair share, on the assumption that there is little 1−
k=1
0
Ik
0
difference between the (N − M ) information already used and equal to c −Nc(r+g)
0 . In the limited service scheme, the allotted
the new (N −M ) information that will arrive later. P2 allocates timeslot Gi must be smaller than Gmax , Gi = c0 Oi = c0 Ii ≤
a timeslot as follows: Gmax . Therefore we can determine si as follows.

 Ri if si = 1 (r+g)
1 if NP
(
k+1
PN N Ii ≤ Gmax
Gi = N G max − s R
j j (2) si = 1− I0 (5)
PNj=1 otherwise k=1 k
0 otherwise

N − sj
j=1
 PN
N Gmax (1−s )+N (r+g)
The cycle time is c = PN i
k=1
because c =
P
1 if min(Rj , Rk ) ≤ N Gmax

where sj = k=1 PN PN
1−
k=1
si I i
PN
0 otherwise k=1 Gk +N (r +g) = c k=1 si Ii +Gmax k=1 (1 − si )+

IEEE/OSA JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. XX, NO. XX, NOVEMBER 2004 4

TABLE II
C OMPARISONS OF FOUR BANDWIDTH ALLOCATION SCHEMES UNDER NON - UNIFORM TRAFFIC CONDITIONS .

Scheme Operation Information used Utilization Fairness


Fixed Gk+1
i = Gmax nothing poor O
Limited Gk+1
i = min(Rik , Gmax ) my own REPORT Ri moderate O
(
Rik if Rik ≤ Gmax
Gk+1
i =
min(Rik , min(Gmax , Fik )) otherwise my own REPORT Ri and
P1 good X
other ONUs’ GATE Gj
P(i−1) mod N
where Fik = N Gmax − j=(i−N +1) mod N
Gj

 Ri if si = 1
PN
Gk+1
i = N Gmax − sj R j
 PNj=1 otherwise
N − sj my own REPORT Ri and
P2  j=1 good O
N
P other ONUs’ REPORT Rj
1 if min(Rj , Rk ) ≤ N Gmax

where sj = k=1

0 otherwise

N (r + g). The output rate can be calculated as 4) Utilization comparison under uniform and non-uniform
 traffic conditions: To compare utilization of the fixed, limited
 Ii PN if si = 1 and proposed schemes under uniform and non-uniform traffic
Oi = Gmax (1− si I i ) (6) conditions, we choose the traffic rates of ONUs as ONU1∼30 :
 c = PN k=1
N (r+g) otherwise
(1−si )+
k=1 Gmax
ONU31∼32 = 1 : k where k is the non-uniformness parameter.
By using the cycle time and output rate, we can obtain the Fig. 4 shows the utilization of the four schemes with k =1,
mean granted timeslot size for ONUi , Gi = cOi and the stable 10, 100.
condition (si = 1, ∀ i) is given by Under uniform traffic conditions, that is when k=1, the four
schemes have equal utilization. In fact, the proposed schemes
N (r + g)
PN Ii ≤ Gmax f or all i. (7) perform almost the same as the limited service scheme in this
1 − k=1 Ik case. Note that the utilization of the conventional schemes
3) Proposed schemes 1 & 2: The total utilizations of P1 decreases as non-uniformness k increases. However, the pro-
and P2 are equal at the steady state even though Oi may be posed schemes make it possible to achieve the same utilization
different from each other. Therefore we will consider only as under uniform traffic conditions.
P2
PNin the following analysis. For every i, we set si = 1 if
N Gmax
k=1 min(I i , I k ) ≤ N Gmax +N (r+g) , otherwise si = 0. If B. Maximum Utilization without considering the RTT effect
there exists i such that si 6= 0, at least one ONUi is not
satisfied even though OLT grants total N Gmax per cycle. In Now, let us find the maximum utilization when the traffic
this case, the cycle time will be c = N Gmax + N (r + g). load offered by the ONUs, except for ONU1 , is 0 ≤ ρ ≤
(r+g) (N −1)Gmax
Otherwise, c = NP N because the total input rate is equal N Gmax +N (r+g) . This can be done by putting (10) into the
1− Ik
c−N (r+g)
k=1 results of the previous subsection.
to c . The output rate can be calculated as
N
X

 Ii PN if si = 1 I1 = 1 − ρ and Ii = ρ, Ii = Oi for i 6= 1
N Gmax
Oi = N Gmax +N (r+g)
− sk I k (8) i=2 (10)
 NP k=1
otherwise N
N− sk X
k=1
U= Oi = O1 + ρ and s1 = 0, si = 1, for i 6= 1
i=1
By using the cycle time and output rate, we can obtain mean
granted timeslot size for ONUi , Gi = cOi and stable condition The maximum utilization of the fixed, limited and proposed
is given by schemes can be obtained as follows:
N
N Gmax Gmax
UF ixed = +ρ (11)
X
Ik ≤ . (9) N Gmax + N g
N Gmax + N (r + g)
k=1

Note that (9) does not have any condition on i. It requires Gmax + ρN (r + g)
ULimited = (12)
only that the total input rate is smaller than the effective Gmax + N (r + g)
N Gmax
capacity Cef f = N Gmax +N (r+g) . This distinguishing feature
allows the proposed schemes to provide full utilization, even N Gmax
UP roposed = (13)
under non-uniform traffic conditions. N Gmax + N (r + g)
IEEE/OSA JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. XX, NO. XX, NOVEMBER 2004 5

C. Maximum Utilization, taking into account the RTT effect V. S IMULATION R ESULTS
In EPON, the geographical distance from an OLT to ONUs In this section, we present simulation results to verify the
can be from several kilometers to 20km [4] and the round- analysis and demonstrate the performance of the proposed
trip time is not negligible. As seen in Fig. 5, the RTT may schemes. Fig. 7 shows our simulation model using an OP-
result in waste of timeslots if condition (14) is satisfied. This NET simulator [12]. We generate the self-similar traffic by
is because the OLT must wait for a previous REPORT to send aggregating 32 pareto-distributed ON-OFF sources [13], [14].
a new GATE. We generate Ethernet frames exponentially distributed with
X
Gk + (N − 1) r + N g < RT Ti (14) mean 500bytes from 64bytes to 1518bytes. Table III shows
k6=i the parameters used in this simulation.
In this case, maximum utilization is smaller than derivations TABLE III
(11)∼(13); thus we need a modification. In this subsection, we S IMULATION PARAMETERS
will take the RTT into consideration and derive the maximum
utilization as the load offered ρ by ONUs, except ONU1 . EPON Link rate (BW ) 1Gbps
Number of ONUs (N ) 8, 16, 32
1) Fixed service scheme: The fixed service scheme is Maximum timeslot (Gmax ) 125 µs
not effected by the RTT because it does not use REPORT. Control message length (r) 0.512µs (=64bytes)
Therefore, the maximum utilization is the same as in the Guard time (g) 1, 5, 10µs
previous derivation (11). Round trip time (RTT) 50,100,200µs
OLT/ONU queue size 300kbytes
Gmax
UF ixed = +ρ (15)
N Gmax + N g
2) Limited service scheme: Maximum utilization will be Fig. 8 depicts the maximum utilization versus network load
achieved when ONU1 uses Gmax per cycle. To find the min- (i.e., the offered load by the ONUs except ONU1 ). The fixed
imum offered load ρth to avoid the RTT effect, we formulate service scheme grants the fixed timeslot Gmax to every ONU
the following equations from Fig. 6(a). even if the network load is low. As a result, the maximum
RT T = cρth + (N − 1)r + N g (16) utilization is very low. The limited service scheme can achieve
higher maximum utilization than the fixed service scheme
c = Gmax + r + RT T (17) because ONU1 gets the chance to send as much data as
By using (16), (17), we can obtain ρth as follows. the other ONUs do not use at an earlier time. However, the
  maximum utilization of the limited service scheme decreases
RT T − (N − 1)r − N g as the non-uniformness increases (i.e. network load decreases),
ρth = max , 0 (18)
Gmax + r + RT T and the RTT effect prevents high maximum utilization from
When ρ ≤ ρth , the cycle time is Gmax + r + RT T . Therefore, being achieved in the low offered load region ( ρ < ρth ). For
the maximum utilization is modified in the region of ρ ≤ ρth . example, maximum utilization is very low in ρ < 0.34 when

Gmax
RT T =100µs, g=1µs, N =16. If we increase RT T to 200µs,
 Gmax +r+RT T + ρ if ρ ≤ ρth
 the situation becomes more serious.
ULimited = (19) Note that the proposed schemes can always achieve almost
 Gmax +ρN (r+g)

if ρ > ρth
Gmax +N (r+g) full utilization, regardless of the network load. In addition, the
3) Proposed scheme 1 & 2: Maximum utilization will be maximum utilization of conventional schemes decreases as the
achieved when the ONUs use N Gmax per cycle. To find number of ONUs and guard time increases, while the proposed
the minimum offered load ρth to avoid the RTT effect, we schemes are almost independent of the two variables.
formulate the following equations from Fig. 6(b). To observe the fairness performance of the four schemes, we
construct a simulation scenario as follows. ONU1 sends traffic
RT T = cρth + (N − 1)r + N g (20) at 300Mbps from 0sec and ONU2 starts to send 300Mbps
c = N Gmax + N (g + r) (21) at 10sec under 500Mbps background traffic produced by
ONU3∼16 . Fig. 9 compares the throughput of ONU1 and
By using (20), (21), we can obtain ρth as follows. ONU2 . Until 10sec, ONU1 sends only 60Mbps and 260Mbps,
respectively, in the fixed and limited service schemes. How-
 
RT T − (N − 1)r − N g
ρth = max , 0 (22) ever, both P1 and P2 are allowed to send 300Mbps all. After
N Gmax + N (r + g)
10sec, the throughputs of ONU1 and ONU2 are 60Mbps
+r+RT T
When ρ ≤ ρth ,Pthe cycle time is c = N Gmax1+ρ because in the fixed service scheme. In the limited service scheme,
c = N Gmax − k6=1 Gk +r+RT T = N Gmax −cρ+r+RT T . ONU1 and ONU2 get 180Mbps equally. ONU1 and ONU2
Therefore, the maximum utilization is modified in the region use 195+255=450Mbps in P1, which is 90Mbps higher than
of ρ ≤ ρth . 180+180=360Mbps in the limited service scheme, but the dis-
tribution is unfair. The total utilization of P2 is the same as that

N Gmax (1+ρ)
 N Gmax +r+RT T if ρ ≤ ρth

of P1. However, ONU1 and ONU2 use 225+255=450Mbps
UP roposed = (23)
N Gmax fairly in P2. This means that we achieve both higher utilization
if ρ > ρ


N Gmax +N (r+g) th
and fairness using our P2 scheme.
IEEE/OSA JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. XX, NO. XX, NOVEMBER 2004 6

TABLE IV
Kyuho Son received his B.S. and M.S. degrees
FAIRNESS S IMULATION SCENARIO AND RESULTS in Electrical Engineering and Computer Science
from the Korea Advanced Institute of Science and
Arr. Dept. Input Rate Output Rate (Mbps) Technology (KAIST), Daejeon, Korea in 2002 and
(sec) (sec) (Mbps) Fixed Limited P1 P2 PLACE 2004, respectively. He is currently workging toward
ONU1 0 ∞ 300 60 180 195 225 PHOTO the Ph.D. degree in Electrical Engineering and Com-
HERE puter Science at the same place. His current research
ONU2 10 ∞ 300 60 180 255 225
interests are in the areas of bandwidth allocation for
ethernet passive optical networks and radio resource
management for 3G/4G wireless network systems.
VI. C ONCLUSION
In this paper, we point out the low utilization problem
of conventional bandwidth allocation schemes under non-
uniform traffic conditions. To resolve this problem, we propose
new schemes that relax the maximum timeslot restriction
and make an intelligent decision about bandwidth allocation
using other ONUs’ information. Our proposed schemes always
utilize more than 90% of bandwidth under any circumstances.
P1, using previous GATE information, may fail to guarantee
fairness. However, P2, using previous REPORT information,
Hyung-Keun Ryu received the B.S. degree in
guarantees max-min fairness as well. Based on our analysis Electronic Engineering from Kyungpook National
and simulation, we conclude that our proposed scheme (P2) University, Daegu, Korea and the M.S. degree in
always fully utilizes the bandwidth under non-uniform traffic Electrical Engineering and Computer Science from
PLACE Korea Advanced Institute of Science and Technology
conditions and guarantees max-min fairness. PHOTO (KAIST), Daejeon where his is currently pursuing
HERE the Ph.D. degree. Since 1991, he has been a Re-
searcher in Research and Development Laboratory,
R EFERENCES Korea Telecom (KT), Korea, where he engaged in
[1] G. Pessavento and M. Kelsey, ”PONs for the Broadband Local Loop,” the areas of Broadband Network and Next Genera-
Lightwave, vol.16, no.10, pp.68-74, Sep. 1999. tion Network. His current research interests include
[2] B. Lund, ”PON Architecture ’futureproofs’ FTTH,” Lightwave, vol.16, aggregate flow control, edge-based QoS provisioning.
no.10, pp.104-107, Sep. 1999.
[3] G. Kramer, G. Pessavento, ”Ethernet Passive Optical Network (EPON):
Building a Next-Generation Optical Access Network,” IEEE Communi-
cation Magazine, vol.40, issue 2, pp.66-73, Feb. 2002.
[4] IEEE Draft P802.3ahTM/D1.2, ”Media Access Control Parameters, Phys-
ical Layers and Management Parameters for subscriber access networks,”
Dec. 2002, http://grouper.ieee.org/groups/802/3/efm.
[5] G. Kramer, B. Mukherjee, and G. Pessavento, ”Ethernet PON (ePON):
Design and Analysis of an Optical Access Newtork,” Photonic Network
Communications, vol.3, no.3, pp.307-319, Jul. 2001.
[6] G. Kramer, B. Mukherjee, and G. Pessavento, ”IPACT: A Dynamic
Protocol for an Ethernet PON (EPON),” IEEE Communication Magazine,
vol.40, issue 2, pp.74-80, Feb. 2002. Chong Song received his B.S. and M.S. degrees in
[7] O. Yoshihara, Y. Fujimoto, N. Oota, N. Miki, ”High Performance EPON,” Control and Instrumentation Engineering from the
IEEE 802.3ah Ethernet in the First Mile Task Force, Nov. 2001, [On- Seoul National University, Seoul, Korea, in 1988 and
line]. Available at http://grouper.ieee.org/groups/802/3/efm/public/nov01/ 1990, respectively, and his Ph.D. degree in Electrical
yoshihara 1 1101.pdf. PLACE and Computer Engineering from the University of
[8] G. Kramer, B. Mukherjee, S. Dixit, Y. Ye, and R. Hirth, ”Supporting PHOTO Texas at Austin in 1995. From 1994 to 1996, he
differentiated classes of service in Ethernet passive optical networks,” HERE was a Member of Technical Staff in the Performance
OSA Journal of Optical Networking, vol.1, no.8&9, pp.280-298, Aug. Analysis Department at AT&T Bell Laboratories,
2002. Holmdel, New Jersey, USA. He is currently an
[9] Maode Ma, Yongqing Zhu, and Tee Hiang Cheng, ”A Bandwidth Guar- Associate Professor with the Department of Elec-
anteed Polling MAC Protocol for Ethernet Passive Optical Networks,” trical Engineering and Computer Science, Korea
Proceedings of IEEE INFOCOM 2003, pp.22-31, Mar. 2003. Advanced Institute of Science and Technology (KAIST), Daejeon, Korea.
[10] C.M. Assi, Y. Ye, S. Dixit and M.A. Ali, ”Dynamic Bandwidth Al- His research interests are in high-speed communication networks, high-
location for Quality-of-Service Over Ethernet PONs,” IEEE Journal on performance switching/routing systems, multimedia networking and perfor-
Selected Areas in Communications, vol.3, no.9, pp.1467-1477, Nov. 2003. mance evaluation. He has published more than 25 papers in international
[11] Fu-Tai An, Yu-Li Hsueh, Kyeong Soo Kim, Ian M. White, and Leonid journals and conferences and holds three U.S. patents with several others
G. Kazovsky, ”A new dynamic bandwidth allocation protocol with quality pending. He is an Editor and an Associate Publication Editor for the Journal of
of service in Ethernet-based passive optical network,” International Communications and Networks. He has served as a Technical Program Com-
Conference on Wireless and Optical Communication (WOC 2003), Jul. mittee member of IEEE INFOCOM (‘97, ‘99, ‘03), an Organizing/Program
2003. Committee member of PV (‘01-‘04) and Technical Program Co-chair of ICBN
[12] OPNET Modeler 7.0, http://www.opnet.com. ‘04.
[13] W. Willinger, M. Taqqu, R. Sherman and D. Wilson, ”Self-similarity
through high-variablity: statical analysis of Ethernet LAN traffic at the
source level,” IEEE/ACM Transactions on Netwrking, vol. 5, issue 1,
pp.71-86, Feb. 1997.
[14] W. Leland, M. Taqqu, W. Willinger, and D. Wilson, ”On the self-similar
nature of Ethernet traffic (extended version),” IEEE/ACM Transactions on
Netwrking, vol. 2, issue 1, pp.1-15, Feb. 1994.
IEEE/OSA JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. XX, NO. XX, NOVEMBER 2004 7

Bandwidth GATE
allocation 1 N N 1 2 1 1 User
scheme 1 1
1 1 1 1
REPORT Splitter/
Coupler
1 N N 1 2 1 N N 1 2 2 User
2 2
1 1 2 NN 2 2

OLT
guard time
1 N N 1 2 N N User
N N
NN N N
ONUs
Fig. 1. Typical topology of EPON

1. Receive REPORT from ONUi 2. Grant the timeslot

Rx 1 R g 2 R N R g 1 R g
OLT
Tx G
4. Send GATE to ONUi
now SEI new SEI

Rx 3. Update SEI
G
ONUi
Tx 1 R

(a) In the case of SEI now + RTT1

1. Receive REPORT from ONUi 2. Grant the timeslot


Waste!
Rx 1 R g 2 R N R g 1 R g
OLT
Tx G
4. Send GATE to ONUi
now SEI new SEI

Rx 3. Update SEI
G
ONUi
Tx 1 R

RTT1
(b) In the case of SEI < now + RTT1

i Data from ONUi G GATE message R REPORT message g guard time

Fig. 2. GATE scheduling: (a) In the case of SEI ≥ now + RT Ti , (b) In the case of SEI < now + RT Ti
IEEE/OSA JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. XX, NO. XX, NOVEMBER 2004 8

New coming ONU4


Limited by Gmax
1st
1 R1=inf g 2, waste R2=0 g 3, waste R3=0 g 4, waste R4=inf g
cycle

Limited by Gmax Limited by Gmax


2nd
1 R1=inf g 2, waste R2=0 g 3, waste R3=0 g 4 R4=inf g Fair
cycle

(a) Fixed scheme

New coming ONU4


Limited by Gmax
1st
1 R1=inf g R2=0 g R3=0 g R4=inf g
cycle

Limited by Gmax Limited by Gmax


2nd
1 R1=inf g R2=0 g R3=0 g 4 R4=inf g Fair
cycle

(b) Limited scheme

New coming ONU4


F1=4Gmax
1st
1 R1=inf g R2=0 g R3=0 g R4=inf g
cycle
F4=0,
F1=4Gmax but minimum guarantee Gmax
2nd
1 R1=inf g R2=0 g R3=0 g 4 R4=inf g
cycle

F1=3Gmax F4=Gmax
3rd
1 R1=inf g R2=0 g R3=0 g 4 R4=inf g Unfair
cycle

(c) Proposed scheme 1

New coming ONU4


s1=0,s2=1,s3=1,s4=1
1st
1 R1=inf g R2=0 g R3=0 g R4=inf g
cycle

s1=0,s2=1,s3=1,s4=0 s1=0,s2=1,s3=1,s4=0
2nd
1 R1=inf g R2=0 g R3=0 g 4 R4=inf g Fair
cycle

(d) Proposed scheme 2

i Data from ONUi R REPORT message g guard time

Fig. 3. Fairness example: (a) Fixed scheme, (b) Limited scheme, (c) Proposed scheme 1, (d) Proposed scheme 2
IEEE/OSA JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. XX, NO. XX, NOVEMBER 2004 9

1
N=32, Gmax = 125us, guard time = 10us
0.9

0.8 Limited, Fixed for k = 1


k=10
Proposed schemes for all k
0.7

0.6
Utilization

Limited k=100
0.5

0.4 k=10

0.3

0.2
Fixed k=100
0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Offered load
Fig. 4. Utilization comparison under uniform and non-uniform traffic conditions

guard time Waste!


i R i+1 R i-1 R i
OLT
G

ONUi
RTTi
i Data from ONUi G GATE message R REPORT message

Fig. 5. RTT effect: timeslot waste due to round-trip time


IEEE/OSA JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. XX, NO. XX, NOVEMBER 2004 10

c th
guard time
Gmax
1 R 2 R 3 R N R 1
OLT
G

ONU1
c RTT

(a)
NGmax c th

1 R 2 R 3 R N R 1
OLT
G

ONU1
c RTT

(b)
Fig. 6. ρth calculation (a) Limited service scheme, (b) Proposed schemes

  )+*-, .0/1(2 3 2 4 5-6



7 6859:.<; =
1?>A@'6='*
tr s!uv s w vyxyz s w v vIy‚ ƒA„†… ‡0ˆ‰(Š ‹ Œ … Ž

e{ | |~} rI s €  Žm(‡0‡<Œ 
 Žm”
§L¨0©+ª«ªU¬ ­    !  ‘?’A“
® ¬¯I° B'CEDGFIH "$#%&' (

LJ KNMAOAPRQ OISUTWVNX X YGZ+VS[Q Y\M^]Z$T_N`a_I]   •+–†— ˜š™›&œ  ž — Ÿ- 


 ¡  mŸ(˜0˜šž ¢
bced f'gAhLijAd k d lmg+h\inpoAiAn$q £?¤A¥  m¢¦
Fig. 7. Simulation model using OPNET
IEEE/OSA JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. XX, NO. XX, NOVEMBER 2004 11

RTT = 50us & guard time = 1us RTT = 50us & guard time = 5us RTT = 50us & guard time = 10us
1 1 1

0.9 0.9 0.9

0.8 0.8 0.8

Maximum Uitlization
Maximum Uitlization
Maximum Uitlization

0.7 0.7 0.7

0.6 0.6 0.6


P 1 (N=8) P 1 (N=8) P 1 (N=8)
0.5 P 1 (N=16) 0.5 P 1 (N=16) 0.5 P 1 (N=16)
P 1 (N=32) P 1 (N=32) P 1 (N=32)
0.4 P 2 (N=8) 0.4 P 2 (N=8) 0.4 P 2 (N=8)
P 2 (N=16) P 2 (N=16) P 2 (N=16)
P 2 (N=32) P 2 (N=32) P 2 (N=32)
0.3 0.3 0.3
Limited (N=8) Limited (N=8) Limited (N=8)
Limited (N=16) Limited (N=16) Limited (N=16)
0.2 Limited (N=32)
0.2 Limited (N=32)
0.2 Limited (N=32)
Fixed (N=8) Fixed (N=8) Fixed (N=8)
0.1 0.1 0.1

± ± ±
Fixed (N=16) Fixed (N=16) Fixed (N=16)
Fixed (N=32) Fixed (N=32) Fixed (N=32)
0 0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Offered load by ONUs except ONU Offered load by ONUs except ONU Offered load by ONUs except ONU
RTT = 100us & guard time = 1us RTT = 100us & guard time = 5us RTT = 100us & guard time = 10us
1 1 1

0.9 0.9 0.9

0.8 0.8 0.8

Maximum Uitlization
Maximum Uitlization
Maximum Uitlization

0.7 0.7 0.7

0.6 0.6 0.6


P 1 (N=8) P 1 (N=8) P 1 (N=8)
0.5 P 1 (N=16) 0.5 P 1 (N=16) 0.5 P 1 (N=16)
P 1 (N=32) P 1 (N=32) P 1 (N=32)
0.4 P 2 (N=8) 0.4 P 2 (N=8) 0.4 P 2 (N=8)
P 2 (N=16) P 2 (N=16) P 2 (N=16)
P 2 (N=32) P 2 (N=32) 0.3 P 2 (N=32)
0.3 0.3
Limited (N=8) Limited (N=8) Limited (N=8)
Limited (N=16) Limited (N=16) Limited (N=16)
0.2 0.2 0.2
Limited (N=32) Limited (N=32) Limited (N=32)
Fixed (N=8) Fixed (N=8) Fixed (N=8)
0.1 0.1 0.1

± ± ±
Fixed (N=16) Fixed (N=16) Fixed (N=16)
Fixed (N=32) Fixed (N=32) Fixed (N=32)
0 0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Offered load by ONUs except ONU Offered load by ONUs except ONU Offered load by ONUs except ONU
RTT = 200us & guard time = 1us RTT = 200us & guard time = 5us RTT = 200us & guard time = 10us
1 1 1

0.9 0.9 0.9

0.8 0.8 0.8


Maximum Uitlization
Maximum Uitlization
Maximum Uitlization

0.7 0.7 0.7

0.6 0.6 0.6


P 1 (N=8) P 1 (N=8) P 1 (N=8)
0.5 P 1 (N=16) 0.5 P 1 (N=16) 0.5 P 1 (N=16)
P 1 (N=32) P 1 (N=32) P 1 (N=32)
0.4 P 2 (N=8) 0.4 P 2 (N=8) 0.4 P 2 (N=8)
P 2 (N=16) P 2 (N=16) P 2 (N=16)
0.3 P 2 (N=32) 0.3 P 2 (N=32) 0.3 P 2 (N=32)
Limited (N=8) Limited (N=8) Limited (N=8)
Limited (N=16) Limited (N=16) Limited (N=16)
0.2 0.2 0.2
Limited (N=32) Limited (N=32) Limited (N=32)
Fixed (N=8) Fixed (N=8) Fixed (N=8)
0.1 0.1 0.1

± ± ±
Fixed (N=16) Fixed (N=16) Fixed (N=16)
Fixed (N=32) Fixed (N=32) Fixed (N=32)
0 0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Offered load by ONUs except ONU Offered load by ONUs except ONU Offered load by ONUs except ONU

Fig. 8. Maximum utilization of different bandwidth allocation schemes versus the offered load by ONUs except ONU 1 , with the number of ONUs (N )
(Line: Analysis, Symbol: Simulation)
IEEE/OSA JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. XX, NO. XX, NOVEMBER 2004 12

N = 16, guard time = 5us


350
Fixed ONU1 Fixed ONU2
Limited ONU1 Limited ONU2
300 P1 ONU1 P1 ONU2
P2 ONU1 P2 ONU2

250
Throughput (Mbps)

200

150

100

50

0
0 5 10 15 20 25
Time (sec)
Fig. 9. Comparison of throughput of ONU1 and ONU2

Das könnte Ihnen auch gefallen