![](https://csdnimg.cn/release/download_crawler_static/16431869/bg2.jpg)
1045-9219 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPDS.2014.2347031, IEEE Transactions on Parallel and Distributed Systems
IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, VOL. XX, NO. X, 2014 2
bution of real-time traffic to achieve excellent QoS still
remains problematic.
Distinct from previous studies, we present in this
paper a load distribution model to optimize the goodput
[12] performance of real-time traffic over multipath net-
works. Goodput differs from throughput as it represents
the amount of data successfully received by the destina-
tion within the imposed deadline. The path diversity and
unreliability in heterogeneous overlay networks [13], in
concert with the stringent QoS requirements, pose crucial
challenges to achieve the goal. To effectively aggregate
the available capacity of different network paths, we
have to seriously consider: (1) how to guarantee the
input traffic delivering within the delay constraint, and
(2) how to alleviate the burst data losses
1
frequently en-
countered in wired/wireless packet switching networks.
Motivated by addressing the above issues, we propose
a Goodput-Aware Load distribuTiON model (GALTON)
that includes three phases: (1) path status estimation
to capture the physical characteristics of each transport
link, (2) flow rate assignment to optimize the aggregate
goodput of input traffic, and (3) deadline-constrained
packet interleaving to alleviate consecutive losses. The
detailed descriptions of the proposed solution will be
presented in Section 4.
Specifically, the contributions of this paper can be
summarized as follows:
• A load distribution model that effectively integrates
the path status estimation, flow rate assignment, and
packet interleaving to optimize the goodput perfor-
mance of real-time traffic over multipath networks.
• A mathematical formulation for load distribution
of multiple deadline-constrained flows over parallel
communication paths to maximize the aggregate
goodput. The utility theory is employed to derive
the solution for flow rate assignment.
• Extensive semi-physical emulations in Exata involv-
ing both real Internet traffic and H.264 video stream-
ing over wired/wireless multipath networks. Exper-
imental results show that: (1) GALTON improves
the goodput by up to 0.25, 0.48, and 0.64 Mbps
compared to the OPI [2], E-DCLD [1], and THR
[45], respectively. (2) GALTON increases the average
video PSNR by up to 6.1, 8.6, and 12.1 dB com-
pared to the OPI, E-DCLD, and THR, respectively.
(3) GALTON reduces the average end-to-end delay
by up to 33.1, 12.3, and 46.5 ms compared to the
OPI, E-DCLD, and THR, respectively. (4) GALTON
mitigates the aggregate loss rates by up to 5.1%,
9.5%, and 10.9% compared to the OPI, E-DCLD, and
THR, respectively. Furthermore, the superiority of
1. In the context of heterogeneous overlay networks, the packet
losses can be classified into three categories: 1) the losses caused by
congestions due to the bandwidth limitation or buffer overflow; 2) the
errors caused by noise or interference in the wireless networks; and
3) the path failure loss or handover loss. In wireless networks, most
packet losses are due to the wireless channel fluctuations or path failure
and not caused by the link congestion.
GALTON over the competing schemes become more
obvious as the number of available paths increases.
The remainder of this paper is structured as follows.
In Section 2, we briefly review and discuss the related
work. The system model and problem formulation are
presented in Section 3. Section 4 describes the solution
procedure of the proposed GALTON. Performance eval-
uation is provided in Section 5 and concluding remarks
are given in Section 6. The basic notations used through-
out this paper are presented in Table 1.
TABLE 1
Basic notations used throughout this paper.
Symbol Definition
P, E the probability value, expectation value.
P, p the set of available paths, a path element.
F, f the set of traffic flows, a flow element.
P, F the number of available paths, traffic flows.
R, R
f
p
the flow rate assignment matrix, an element.
T the delay constraint for the input traffic.
RT T
p
the round trip time of p.
µ
p
, ν
p
the available, residual bandwidth of p.
G/B the Good/Bad state of p.
π
B
p
, π
G
p
the stationary probability that p is in B/G state.
ξ
G
p
the state transition probability of p from B to G.
π
p
the transmission loss rate of p.
Π
f
p
the effective packet loss rate of flow f over p.
R
p
, P kt
p
the probing traffic rate/size of p.
M, M
p
the total number of packets, dispatched onto p.
D
f
p
the end-to-end delay of flow f over path p.
Θ the aggregate goodput.
U, U
f
p
the system utility matrix, an element.
2 RELATED WORK
Traffic load distribution over multipath networks has
been an active research area in recent years and the
general reviews can be found in [9][14]. The existing
distribution models can be categorized into the flow and
packet based traffic splitting approaches.
The packet based models generally dispatch packets
onto different paths based on the channel status in-
formation, packet delay constraint, etc. Although these
scheduling policies are able to reduce the queueing
delay over a single path, they may also induce serious
packet reordering problems, which in turn result in
large end-to-end delay. The Effective Delay Controlled
Load Distribution (E-DCLD) [1] aims at minimizing the
difference among end-to-end delays of different paths,
thereby reducing packet delay variation and risk of pack-
et reordering. The authors formulate the queueing delay
for each path with a hybrid M/M/1 model and splits
the traffic to minimize the cost of delay variations. In
[18], each path is assigned with a weight associated with
评论0
最新资源