Hindawi Publishing Corporation

International Journal of Aerospace Engineering

Volume 2015, Article ID 329820, 9 pages
http://dx.doi.org/10.1155/2015/329820

Research Article
Optimal Configuration of Virtual Links for
Avionics Network Systems

Dongha An, Kyong Hoon Kim, and Ki-Il Kim

Department of Informatics, Research Center for Aerospace Parts Technology, Gyeongsang National University,
Jinju-daero 501, Jinju 660-701, Republic of Korea

Correspondence should be addressed to Kyong Hoon Kim; [email protected]

Received 28 June 2015; Accepted 18 October 2015

Academic Editor: Linda L. Vahala

Copyright © 2015 Dongha An et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

As the bandwidth and scalability constraints become important design concerns in airborne networks, a new technology, called
Avionics Full Duplex Switched Ethernet (AFDX), has been introduced and standardized as a part 7 in ARNIC 664. However,
since previous research interests for AFDX are mainly bounded for analyzing the response time where flows information is given,
configuration problem for both Maximum Transmission Unit (MTU) and Bandwidth Allocation Gap (BAG) over virtual links in
AFDX networks has not been addressed yet even though it has great impact on required bandwidth. Thus, in this paper, we present
two configuration approaches to set MTU and BAG values on virtual links efficiently while meeting the requirement of AFDX.
The first is to search available feasible configuration (MTU, BAG) pairs to satisfy application requirements as well as AFDX switch
constraints, and the second is to get an optimal pair to minimize required bandwidth through well-known branch-and-bound
algorithm. We analyze the complexity of the proposed algorithm and then evaluate the proposed algorithm by simulation. Finally,
we prove that the proposed schemes are superior to general approach in the aspects of speed and required bandwidth in AFDX
networks.

1. Introduction are introduced and implemented in AFDX. One is traffic

control by guaranteeing the bandwidth of each application,
Due to increasing demands in Aircraft Data Networks and the other is duplicated transmission for reliability along
(ADNs), such as high available bandwidth, minimum wiring dual redundant channels. While the former aims to bound
to reduce the weight, and low development cost, typical com- the jitter and transmit latency within the deadline, the
munications technologies are expected to be replaced by new latter transmits the same data stream over disjoint networks
one. In other words, since ARNIC 429, MIL-STD-1553, and concurrently. To achieve this goal, virtual links are introduced
ARNIC 629 cannot accommodate the mentioned demands and created for each (source, destination) pair in AFDX. By
completely, a new technology, called Avionics Full Duplex controlling these virtual links, deterministic behaviors are
Switched Ethernet (AFDX), has been implemented and then completely guaranteed. Thus, determining flows for the same
standardized for new ADN [1–3]. As a new technology for virtual link and its properties become the network designer’s
ADN, reliable and deterministic property as well as imple- great task when it comes to configuring AFDX networks.
mentation cost should be considered. As a result, AFDX was Two parameters of virtual links in AFDEX networks are
introduced and recently deployed in data networks on the important in terms of real-time requirements: BAG (Band-
Airbus 380 aircraft. width Allocation Gap) and MTU (Maximum Transfer Unit).
In the point of technical issue, the AFDX follows original BAG is the time duration or period between two consecutive
Ethernet for compatibility and scalability properties and frames, where the value of BAG is 1 and 128 msec in a form of
extends it to ensure deterministic behavior and high relia- power of 2. MTU is the size of message in each frame. Thus,
bility. In order to comply with the stringent requirements the AFDX configuration problem is to determine these two
of ADNs as well as ensuring them, two new functions parameters of each virtual link without missing the deadline.
2 International Journal of Aerospace Engineering

In [4], they provided several algorithms and integer- That is, a message of 𝑙𝑖 bytes is generated every 𝑝𝑖 time units
linear programming formulation in order to solve the trans- and is delivered to the destination application.
mission parameters of virtual links as well as data routing The basic communication unit in AFDEX networks is
over links in the AFDEX networks. They provided optimal defined as a virtual link (VL). For example, Figure 1 shows
solutions to minimize reserved bandwidth as well as actually three virtual links among LRUs. These virtual links sharing
consumed bandwidth, respectively. In [5], modeling method physical links are scheduled in AFDX network switches.
for AFDX frame management was introduced to ascertain the Furthermore, multiple applications transmit real-time mes-
reliability properties of design. Another important property sages throughout a common virtual link if their source and
of AFDX, reliability through redundant transmissions, was destination units are the same. In the example of Figure 1, two
analyzed by formal method in [6]. application messages are shared in the virtual link VL3 .
Another related work is to analyze the system metric such There are two important parameters in a virtual link. The
as response time of AFDX networks. In [7], they applied first is Bandwidth Allocation Gap (BAG) to specify a periodic
network calculus, queuing networks simulation and model frame. In AFDX switches, a BAG is defined by a value of
checking into evaluating bounding end-to-end delays on 2𝑘 msec, where 𝑘 = 0, 1, . . . , 7. As all BAGs are 2𝑘 msec,
AFDX networks. In [8], they showed that Trajectory approach virtual links are multiplexed in AFDX switches. The second
which analyzes the worst-case delays throughout message parameter is Maximum Transfer Unit (MTU) of the message
flows outperforms the network calculus method under indus- in bytes at each frame. Payloads of applications in a virtual
trial configuration. In [9], they obtain worst-case latency and link are transmitted within maximum MTU bytes in a single
output jitter for the network messages by defining a real- frame. If the size of a payload is greater than the MTU, it is
time model for a communications network based on AFDX. fragmented into multiple frames. Therefore, a virtual link VL𝑖
Another approach in [10] is performance evaluation system is defined by (BAG𝑖 , MTU𝑖 , 𝐹𝑖 ) as follows:
through simulation with popular NS-2 and analysis of impact
of several system parameters such as scheduling algorithm. (i) BAG𝑖 : Bandwidth Allocation Gap or period of VL𝑖 in
Since most of existing research work has assumed the a value of 2𝑘 msec where 𝑘 = 0, 1, . . . , 7.
preconfigured networks in advance, configuration problem (ii) MTU𝑖 : maximum transfer unit or message size of VL𝑖
should be mentioned prior to deployment. To achieve this in bytes.
goal, in this paper, we focus on two main problems: (i)
finding feasible BAG and MTU parameters of virtual links (iii) 𝐹𝑖 : a set of message flows in VL𝑖 , where the 𝑗th
in an AFDX switch for given virtual links of messages message flow is denoted as 𝑓𝑖,𝑗 = (𝑙𝑖,𝑗 , 𝑝𝑖,𝑗 ).
and (ii) providing an optimal solution in feasible pairs to
minimize the total bandwidth. We define two problems 2.2. Problem Definition. For a given virtual link VL𝑖 , MTU
formally and then solve the problems using the branch-and- and BAG are configured so as to meet all the real-time
bound technique. While our earlier work [11] only provided requirements of message flows in the link. If the payload of
an algorithm to solve the first problem, this paper improves a message is greater than the MTU size, it is transmitted
and extends it to include optimal solution for the purpose of in multiple fragmented packets. Since all BAGs of VLs are
minimizing the total bandwidth. Finally, complexity of the harmonic, the schedulability analysis is easily derived by
proposed scheme and experimental results through diverse utilization analysis. Thus, (1) tells the message constraint of
simulation scenarios are given to demonstrate the suitability VL𝑖 with 𝑛𝑖 messages to guarantee the real-time requirement
of the proposed scheme. of all message flows in the link [2]:
The rest of this paper is organized as follows. The system
𝑛𝑖 ⌈𝑙𝑖,𝑗 /MTU𝑖 ⌉
model and the problem definition are provided in Section 2. 1
∑ ≤ . (1)
The solution of a feasible configuration problem is dealt with
𝑗=1 𝑝𝑖,𝑗 BAG𝑖
in Sections 3 and 4, while the optimal algorithm is provided
in Section 5. Performance evaluations are shown in Section 6. Let us assume that the system has 𝑁 VLs on an AFDX
Finally, conclusion and further work followed in Section 7. switch with 𝐵 bandwidth in bps. Each VL𝑖 is configured
with (MTU𝑖 , BAG𝑖 ), so that MTU𝑖 bytes are transmitted
every BAG𝑖 msec. In addition, each VL message requires the
2. System Model and Problem Definition
overhead of 67 bytes as shown in Figure 2. Since the total
2.1. System Model. In this paper, we assume that an avionics bandwidth of VLs should not exceed the network bandwidth,
system is composed of many computing or sensing units and the following bandwidth constraint should be met:
AFDX network switches to connect other computing units.
𝑛
Thus, an AFDX message is uniquely defined by UDP source MTU𝑖 + 67
8∑ × 103 ≤ 𝐵. (2)
and destination ports, as shown in Figure 1. Sine we focus on 𝑖=1 BAG 𝑖
real-time AFDX messages, a message flow 𝑓𝑖 is defined by (𝑙𝑖 ,
𝑝𝑖 ), where 𝑙𝑖 is the payload of the message in bytes and 𝑝𝑖 The last constraint of virtual link scheduling is about
is Message Transmit Cycle (MTC) of the message in msec. jitter. The maximum allowed jitter on each virtual link in
International Journal of Aerospace Engineering 3

LRU 1

Partition 1 Partition 2
API API API API
port 1 port 2 port 1 port 2

AFDX AFDX AFDX AFDX

port 1 port 2 port 1 port 2

IP 10 IP 20 IP 10 IP 20
UDP 1 UDP 1 UDP 1 UDP 1

MAC 10

VL1 VL3

VL2
LRU 2 AFDX switch LRU 3
MAC 1 MAC 3 MAC 2

IP 1 IP 3 IP 3 IP 2
UDP 1 UDP 2 UDP 2 UDP 1

AFDX AFDX AFDX AFDX

port 1 port 2 port 2 port 1

API API API API

port 1 port 2 port 2 port 1

Partition 1 Partition 1

Figure 1: An example of virtual links in an AFDX switch.

MAC MAC Type IP UDP AFDX Padding

Preamble SFD SN FCS IFG
destination source IPv4 header header payload 0–16
7 bytes 1 byte 6 bytes 6 bytes 2 bytes 20 bytes 8 bytes 1–17 bytes bytes 1 byte 4 bytes 12 bytes

Overhead (50 bytes) AFDX payload Overhead (17 bytes)

Figure 2: The AFDX frame structure and its overhead.

the ARINC 664 specification requires 500 𝜇sec [2]. Thus, the the AFDX switch. The following tells the formal definition of
following equation tells the jitter constraint, where 40 𝜇sec is the problem.
the typical technological jitter in hardware level to transmit
an Ethernet frame: Definition 1. For a given set of virtual links 𝑉 = {VL𝑖 |
𝑖 = 1, . . . , 𝑁}, the problem of AFDX-CONF is to determine
8 ∑𝑛𝑖=1 (67 + MTU𝑖 ) (BAG𝑖 , MTU𝑖 ) of each VL𝑖 so as to satisfy three constraints of
40 + ≤ 500. (3)
(1), (2), and (3), where BAG𝑖 ∈ {1, 2, 4, 8, 16, 32, 64, 128} and
𝐵
MTU𝑖 ∈ {1, 2, . . . , 1471}.
In this paper, we define two problems related to configu-
ration of virtual links of an AFDX switch. The first problem The second one is to find an optimal configuration of vir-
is to find a feasible configuration of BAG and MTU pairs of tual links for the purpose of minimizing the total bandwidth.
virtual links, which satisfies three constraints of (1), (2), and The bandwidth remaining after reserving real-time flows is
(3). This is useful and important when an administrator finds generally used for non-real-time network traffic. Thus, it is
a feasible configuration of a given set of real-time messages in important to find a configuration with the least bandwidth as
4 International Journal of Aerospace Engineering

Algorithm Find Feasible BAG MTU (VL𝑖 )

(1) 𝑁step ← ⌀
(2) for each message 𝑓𝑖,𝑗 in VL𝑖 do
(3) frag ← ⌈𝑙𝑖,𝑗 /(⌈𝑙𝑖,𝑗 /𝑝𝑖,𝑗 ⌉)⌉
(4) while frag ≥ 1 do
(3) ⌈𝑚 ← 𝑙𝑖,𝑗 /frag⌉
(5) 𝑁step ← 𝑁step ∪ {𝑚}
(6) frag ← frag − 1
(7) endwhile
(8) endfor
(9) for 𝑘 from 0 to 7 do
(10) MTU𝑖,𝑘 ← 0
𝑛𝑖
(11) MTU𝑖,𝑘 ← the least 𝑚 ∈ 𝑁step s.t. ∑𝑗=1 (⌈𝑙𝑖,𝑗 /𝑚⌉/𝑝𝑖,𝑗 ) ≤ 1/2𝑘
(12) endfor

Algorithm 1: Algorithm of feasible BAG and MTU pairs of a VL.

long as the configuration is feasible. The following tells the

second problem definition. 1
1/1 (BAG = 1)

Definition 2. For a given set of virtual links 𝑉 = {VL𝑖 | ⌈80/MTU⌉ ⌈100/MTU⌉

0.8 +
𝑖 = 1, . . . , 𝑁}, the problem of AFDX-BOPT is to find (BAG𝑖 , 10 12
Utilization

MTU𝑖 ) of each VL𝑖 so as to minimize the total bandwidth 0.6

and satisfy three constraints of (1), (2), and (3), where BAG𝑖 ∈ 1/2 (BAG = 2)
{1, 2, 4, 8, 16, 32, 64, 128} and MTU𝑖 ∈ {1, 2, . . . , 1471}.
0.4
We solve both problems of AFDX-CONF and AFDX- 1/4 (BAG = 4)
BOPT in two steps. The first step is to find the list of (BAG𝑖 , 0.2
MTU𝑖 ) which guarantees the schedulability of message flows 1/8 (BAG = 8)
in VL𝑖 . Each (BAG𝑖 , MTU𝑖 ) should be selected such that it 0
satisfies the constraint of (1) (in Section 3). Then, we find 0 20 40 60 80 100 120 140
the solution of given virtual links with consideration of two MTU
constraints of (2) and (3) (in Sections 4 and 5). Figure 3: An example of feasible BAG and MTU of a virtual link.

3. Schedulable BAG and MTU Pairs of a VL

In AFDX networks, multiple applications should send their BAG is greater than or equal to 8 msec. Therefore, the feasible
messages to a virtual link queue in the system, so that solutions (BAG1 , MTU1 ) of VL1 are given by (1, 17), (2, 40),
the scheduling issue for those messages is required. Let and (4, 100).
us consider a virtual link VL1 with two message flows of The pseudocode of Algorithm 1 describes how to obtain
𝑓1,1 (80, 10), 𝑓1,2 (100, 12) as an example. The values of BAG the set of feasible BAG and MTU pairs of a given virtual link
and MTU of VL1 are set to satisfy (1) in order to meet the real- VL𝑖 . The first part of the algorithm gathers all step integers
time requirement of two messages. The left side of (1) is shown at which the utilization function begins a new piecewise
in Figure 3 as a step function, while 1/BAG is also drawn in constant due to the ceiling function. We denote the set of such
the figure for different BAG values. step integers as 𝑁step . For each message 𝑓𝑖,𝑗 , such step points
For a given BAG𝑖 , there exist many MTUs which satisfy are derived and added into 𝑁step (lines 1–8).
(1). For example, when BAG1 = 1, all MTUs can be used if
MTU ≥ 17, as shown in Figure 3. Since a longer MTU size Then, for each 2𝑘 value, we find the minimum MTU
requires more bandwidth and jitter, the smallest value should which satisfies (1) (lines 9–12). We denote MTU𝑖,𝑘 by the fea-
be selected. Thus, MTU1 of the example VL1 is 17 bytes when sible MTU in case of BAG𝑖 = 2𝑘 for a virtual link VL𝑖 . If there
BAG1 is 1 msec. Similarly, MTUs of VL1 for BAGs with 2 msec is no feasible MTU, then MTU𝑖,𝑘 = 0. For a given 𝑛𝑖 flows, the
and 4 msec are given by 40 bytes and 100 bytes in each, as time complexity of Algorithm 1 is 𝑂(𝑛𝑖 ⋅ |𝑁step |) since we have
shown in Figure 3. to find and check the feasibility at each step point of messages.
When the MTU size is greater than the maximum pay-
load size of messages, the required utilization is not changed. 4. Feasible BAG and MTU Pairs of VLs
For example, the lower bound of the utilization of VL1 is given
by about 0.1834 at MTU = 100. This implies that there is no 4.1. Problem Definition. The problem of finding feasible BAG
MTU which guarantees the schedulability of two messages if and MTU pairs of a given set of virtual links is not trivial. For
International Journal of Aerospace Engineering 5

Table 1: An example of virtual links (𝐵 = 1 Mbps).

Flows Payload MTC Feasible BAG and MTU pairs

(𝑓𝑖,𝑗 ) (𝑙𝑖,𝑗 ) (𝑝𝑖,𝑗 ) 𝑠𝑖,𝑘 (bag𝑖,𝑘 , MTU𝑖,𝑘 )
𝑓1,1 200 80
VL1 (1, 5), (2, 9), (4, 17), (8, 34), (16, 67), (32, 200)
𝑓1,2 250 160

VL2 𝑓2,1 250 220 (1, 6), (2, 12), (4, 25), (8, 50), (16, 100), (32, 200)
𝑓2,2 200 40

example, let us consider the example of two virtual links of bandwidth due to overhead of fragmentation. For example,
Table 1 where the network speed (𝐵) is given by 1 Mbps. For if we select (1, 5) and (1, 6) as (BAG, MTU) of two VLs of
each virtual link, the feasible BAG and MTU pairs are derived Table 1, it does not meet the bandwidth constraint of (2).
by Algorithm 1, as shown in the last column of Table 1. Now, On the contrary, if (2, 9) and (2, 12) are selected as (BAG,
a new problem arises about selecting appropriate BAG and MTU) of two VLs, this configuration does not meet the jitter
MTU pairs of two virtual links so as to meet both constraints constraint of (3). The selection of (1, 5) and (2, 12) of VL1 and
of (2) and (3). VL2 satisfies both constraints so that all messages in VLs meet
There are some tradeoffs among feasible BAG and MTU their real-time requirements.
pairs of a virtual link VL𝑖 . Solutions with smaller BAG provide Let us assume that MTU𝑖,𝑘 is derived from Algorithm 1
less jitter due to smaller MTU size, while they require more for each 𝑖 and 𝑘. And, let us denote it by

𝑘
{1 if BAG and MTU of VL𝑖 are set as 2 and MTU𝑖,𝑘 in each,
𝑋𝑖𝑘 = { (4)
0 otherwise.
{

Then, the problem of AFDX-CONF can be stated as the the solutions in the subset satisfy both constraints by using
following linear integer problem with 𝑛 × 8 binary variables the bound function of minimum values of remaining levels.
and three constraints: to find 𝑋𝑖𝑘 The algorithm stops the search of the subset which already
subject to violates one of two constraints.
7 (ii) Branch-and-Bound Strategy. The algorithm searches a
∑ 𝑋𝑖𝑘 = 1 feasible solution in a leaf node in depth-first-search (DFS)
𝑘=0
manner. This algorithm finds a feasible solution when it
𝑛 7
MTU𝑖,𝑘 + 67 reaches any leaf node in the search tree.
𝐵
∑ ( ∑ 𝑋𝑖𝑘 ⋅ )≤ (5)
𝑖=1 𝑘=0 2𝑘 8000 Algorithm 2 finds the set of feasible BAG and
MTU pairs of each virtual link by calling the function
𝑛 7
𝐵 Find Feasible BAG MTU ( ) proposed in Algorithm 1 (lines
∑ ( ∑ 𝑋𝑖𝑘 ⋅ (67 + MTU𝑖,𝑘 )) ≤ 460 ⋅ . 1, 2). Let us denote 𝜎𝑖 by the set of feasible BAG and MTU
𝑖=1 𝑘=0
8
pairs (𝑠𝑖,𝑘 ) of virtual link 𝑖. Then, the bounds of bandwidth
and jitter in virtual link 𝑖 are found by selecting the minimum
4.2. Algorithm. For given 𝑁 virtual links, the exhaustive values (lines 4–7).
search of the problem AFDX-CONF takes 𝑂(8𝑁) since each The function EDFS BandB in Algorithm 2 is the recursive
virtual link might have maximum eight solutions. In this implementation at level 𝑖 in the search tree. Two values of
paper, we provide an enhanced branch-and-bound algorithm 𝐵prev and 𝐽prev are two bounds of bandwidth and jitter of sub-
proposed in [11] in order to find a feasible solution for given 𝑁 solutions from VL𝑖 to VL𝑁. For each 𝑠𝑖,𝑘 = (bag𝑖,𝑘 , MTU𝑖,𝑘 ),
virtual links with their feasible BAG and MTU pairs derived
two constrains of (2) and (3) are checked including a new
by Algorithm 1. In [11], we used the cumulative bandwidth
solution of VL𝑖 (lines 14–16). If either of the two constraints
and jitter for the pruning condition. In this work, however,
we define the bound functions of bandwidth and jitter for is not satisfied, it is pruned. Otherwise, the depth-first-search
the remaining levels in the search tree in order to find is continued with two updated bound values (line 17).
the pruning node as early as possible. The following is the When the search reaches a leaf node, the function returns
pruning condition and branch-and-bound strategy of the true (line 12). The return value of calling EDFS BandB is
proposed algorithm. true; the final solution 𝑆 is updated as to include 𝑠𝑖,𝑘 (line
19) and the function returns true. Thus, the problem of
(i) Pruning Condition. The pruning condition is two con- AFDX-CONF is solved by Algorithm 2. If the return value of
straints of (2) and (3). The algorithm examines whether DFS BandB (𝐵0 , 𝐽0 , 1, 𝑆) is true, a feasible solution is stored
6 International Journal of Aerospace Engineering

𝑛 7
MTU𝑖,𝑘 + 67 𝐵
Algorithm Find Feasible Configurations (𝑉) ∑ ( ∑ 𝑋𝑖𝑘 ⋅ )≤ (8)
2𝑘 8000
/∗ 𝑉 = {VL𝑖 | 𝑖 = 1, . . . , 𝑁} ∗ / 𝑖=1 𝑘=0
(1) for 𝑖 from 1 to 𝑁 do 𝑛 7
(2) call Find Feasible BAG MTU (VL𝑖 ) 𝐵
∑ ( ∑ 𝑋𝑖𝑘 ⋅ (67 + MTU𝑖,𝑘 )) ≤ 460 ⋅ . (9)
(3) 𝑆 ← ⌀ 𝑖=1 𝑘=0
8
(4) for 𝑖 from 1 to 𝑁 do
(5) 𝐵𝑖 ← min𝑠𝑖,𝑘 ∈𝜎𝑖 {(MTU𝑖,𝑘 + 67)/bag𝑖,𝑘 } For example, there are two feasible configurations in
(6) 𝐽𝑖 ← min𝑠𝑖,𝑘 ∈𝜎𝑖 {MTU𝑖,𝑘 + 67} Table 1: {(1, 5), (2, 12)} and {(2, 9), (1, 6)} under the band-
(7) endfor width constraint of 1 Mbps. The first configuration requires
𝑁
(8) 𝐵0 ← ∑𝑖=1 𝐵𝑖 892 kbps, while the second one uses 888 kbps. Thus, in this
𝑁
(9) 𝐽0 ← ∑𝑖=1 𝐽𝑖 case, {(2, 9), (1, 6)} is the optimal configuration of two virtual
(10) result ← EDFS BandB (𝐵0 , 𝐽0 , 1, 𝑆)
links.
(11) return 𝑆
Function EDFS BandB (𝐵prev , 𝐽prev , 𝑖, 𝑆) In order to solve the problem, we provide a branch-and-
(12) if 𝑖 = 𝑁 + 1 then return true bound algorithm. The bound function is the same as one
(13) for each 𝑠𝑖,𝑘 of VL𝑖 do in Algorithm 2, so that we use the minimum bandwidth
(14) 𝐵curr ← 𝐵prev − 𝐵𝑖 + (MTU𝑖,𝑘 + 67)/bag𝑖,𝑘 and jitter for bound values of each node. Thus, the pruning
(15) 𝐽curr ← 𝐽prev − 𝐽𝑖 + MTU𝑖,𝑘 + 67 condition is two constraints of (8) and (9) using the bound
(16) if 𝐵curr ≤ 𝐵/8000 and 𝐽curr ≤ 460 ⋅ 𝐵 then functions. The branch strategy is different from DFS. Since
(17) result ← EDFS BandB (𝐵curr , 𝐽curr , 𝑖 + 1, 𝑆) the algorithm finds an optimal solution as soon as possible,
(18) if result = true then
(19) 𝑆 ← 𝑆 ∪ {𝑠𝑖,𝑘 }
we select a node with the minimum bandwidth bound
(20) return true function and branch its child nodes. The pseudocode of this
(21) endif is shown in Algorithm 3. Since the branch-and-bound tech-
(22) endif nique searches all possible cases in the worst-case, the time
(23) endfor complexity of Algorithm 3 is 𝑂(2𝑁) for given 𝑁 virtual links.
(24) return false In Algorithm 3, a 𝑛𝑜𝑑𝑒 has four data: 𝑛𝑜𝑑𝑒.𝑙𝑒V𝑒𝑙 for the
node level, 𝑛𝑜𝑑𝑒.𝐵 for the bandwidth bound, 𝑛𝑜𝑑𝑒.𝐽 for the
Algorithm 2: The proposed algorithm for AFDX-CONF. jitter bound, and 𝑛𝑜𝑑𝑒.𝑆 for the selected configuration from
the root to the node. We use a priority-queue for managing
live nodes in the order of bandwidth bound. The function
in 𝑆. Otherwise, the empty set is returned, which implies no insert node in Q (𝑖, 𝐵, 𝐽, 𝑆) makes a node with four data
feasible configuration is found for a given set of virtual links. and inserts it into the priority queue. Then, the function
get first in Q ( ) returns the minimum-bandwidth node in
5. Optimal Solution of BAG and MTU Pairs for currently live nodes in the queue. When the algorithm
reaches a leaf node (line 12), it finds an optimal solution and
Reducing Bandwidth
returns the configuration (line 13).
In this section, we focus on finding an optimal configuration The branch strategy is implemented by the priority queue.
of virtual links in terms of the used bandwidth of the con- The function call of get first in Q ( ) selects such node to
figuration. It is important to minimize the total bandwidth branch among live nodes. For all child nodes of the selected
of virtual links for real-time traffic because we can use node, the bound functions of bandwidth and jitter are
the remaining bandwidth for non-real-time network traffic updated (lines 16-17). The pruning condition is checked in
in avionics systems. The problem of AFDX-CONF finds a line 18. If a new child node is not pruned, it is added in the
feasible solution which satisfies three constraints of AFDX priority queue (line 19). The algorithm repeats this procedure
switch as soon as possible, so that it does not minimize the until it finds an optimal solution or there is no live node in the
bandwidth. However, AFDX-BOPT finds a feasible solution queue. If there is no feasible configuration for virtual links, it
which minimizes the total bandwidth, so that the problem is returns the empty set for notifying no feasible solution (line
rewritten as follows: 22).
to minimize
𝑛 7
6. Performance Evaluations
MTU𝑖,𝑘 + 67
8000∑ ( ∑ 𝑋𝑖𝑘 ⋅ ) (6) In this section, we show performance evaluation of the
𝑖=1 𝑘=0 2𝑘
proposed algorithm. First, we evaluate the execution time
of the proposed algorithms compared with the brute-force
subject to search and the previous work [11]. In the experiments, we
7
generate five virtual links with two message flows in each
∑ 𝑋𝑖𝑘 = 1 (7) virtual link. The payload of a message is randomly generated
𝑘=0 from 20 to 80 bytes. The MTC or period of a message is
International Journal of Aerospace Engineering 7

Algorithm Find Minimum Bandwidth Configuration (𝑉)

(1) for 𝑖 from 1 to 𝑁 do
(2) call Find Feasible BAG MTU (VL𝑖 )
(3) for 𝑖 from 1 to 𝑁 do
(4) 𝐵𝑖 ← log𝑠𝑖,𝑘 ∈𝜎𝑖 {(MTU𝑖,𝑘 + 67)/bag𝑖,𝑘 }
(5) 𝐽𝑖 ← log𝑠𝑖,𝑘 ∈𝜎𝑖 {MTU𝑖,𝑘 + 67}
(6) endfor
𝑁
(7) 𝐵0 ← ∑𝑖=1 𝐵𝑖
𝑁
(8) 𝐽0 ← ∑𝑖=1 𝐽𝑖
(9) insert node in 𝑄 (0, 𝐵0 , 𝐽0 , ⌀);
(10) while is empty 𝑄 ( ) = false do
(11) node ← get first in 𝑄 ( );
(12) if node.level = 𝑁 then
(13) return 𝑛𝑜𝑑𝑒.𝑆;
(14) 𝑖 ← node.level + 1
(15) for each 𝑠𝑖,𝑘 of VL𝑖 do
(16) 𝐵curr ← 𝑛𝑜𝑑𝑒.𝐵 − 𝐵𝑖 + (MTU𝑖,𝑘 + 67)/bag𝑖,𝑘
(17) 𝐽curr ← 𝑛𝑜𝑑𝑒.𝐽 − 𝐽𝑖 + MTU𝑖,𝑘 + 67
(18) if 𝐵curr ≤ 𝐵/8000 and 𝐽curr ≤ 460 ⋅ 𝐵 then
(19) insert node in Q (𝑖, 𝐵curr , 𝐽curr , 𝑛𝑜𝑑𝑒.𝑆 ∪ {𝑠𝑖,𝑘 });
(20) endfor
(21) endwhile
(22) return ⌀

Algorithm 3: The Bounded BFS BandB algorithm.

10000 10000
Normalized execution time
Normalized execution time

1000 1000

100 100

10 10

1 1

10~60 60~110 110~160 160~210 210~260 10~60 60~110 110~160 160~210 210~260
MTC MTC

Brute-force Brute-force
DFS B-DFS
EDFS EBFS
(a) AFDX-CONF (b) AFDX-BOPT

Figure 4: Normalized execution times of algorithms.

randomly selected among five different intervals, as shown in As shown in Figure 4, the proposed algorithms run much
Figure 4. The network bandwidth is set as 6 Mbps. faster than the exhaustive search algorithm due to the branch-
For each case of Figure 4, we generate 5000 random sets and-bound technique. Let us note that the 𝑦-axis in Figure 4
of five virtual links and measure the average execution time is log-scale value. Table 2 shows the average number of BAG
of algorithms. Figure 4(a) shows the normalized execution and MTU pairs per virtual link. As the MTCs become larger,
times of the brute-force search, DFS in [11], and EDFS in this the number of possible solutions of each virtual link is
paper for solving AFDX-CONF. Similarly, Figure 4(b) shows increased, which requires more execution time, as shown
the performance of the proposed algorithm EBFS compared in Figure 4. Since we use bound functions for bandwidth
to the brute-force search and the modified version of DFS [11] and jitter, the proposed ones are faster than the previous
for solving AFDX-BOPT. algorithm in [11].
8 International Journal of Aerospace Engineering

40
100000
35
Normalized bandwidth

30
10000

Normalized execution time

25
20
1000
15
10 100
5
0 10
10~60 60~110 110~160 160~210 210~260
MTC
1
Brute-force
B-DFS 1 2 3 4 5 6
EBFS Number of VLs

Figure 5: Bandwidth comparison. BF

DFS
EDFS
Table 2: The average number of BAG and MTU pairs per VL.
(a) AFDX-CONF
MTC 10∼60 60∼110 110∼160 160∼210 210∼260
BAG and MTU pairs 100000
4.7 6.0 6.8 7.0 7.0
per VL
10000
Normalized execution time

Next, we compared the bandwidth reserved by the solu-

tions of AFDX-BOPT with feasible solutions of AFDX-CONF. 1000
As shown in Figure 5, the proposed algorithm for AFDX-
BOPT reduces much bandwidth, so that the remaining band- 100
width can be used for other non-real-time network traffic.
We also analyzed the effect of algorithm execution times
in terms of the number of virtual links. In Figure 6, we 10
increased the number of virtual links from 1 to 6, where each
virtual link includes two messages. As shown in Figure 6, the 1
execution time of brute-force search increases exponentially
1 2 3 4 5 6
since the search space is 𝑂(2𝑁). Let us note that 𝑦-axis in Number of VLs
Figure 6 is log-scale. Although the time complexity of the
proposed algorithm is 𝑂(2𝑁), both EDFS and EBFS are much BF
B-DFS
faster than the previous algorithms in [11]. This is mainly EBFS
because we use efficient bound functions in the algorithms.
(b) AFDX-BOPT
The execution time of EBFS in case of 6 virtual links is
decreased since the algorithm decides that there is no solution Figure 6: Normalized execution times with respect to the number
in the upper-layer of solution trees. of virtual links.

7. Conclusions
In this paper, we provided an efficient algorithm to find Throughout simulations, we showed that the algorithm runs
feasible configuration of an AFDX switch for the purpose fast and reduces much bandwidth compared with the algo-
of meeting the real-time requirements of all messages in rithm to find a feasible set. The proposed algorithm is useful
avionics. Two important parameters of BAG and MTU of when the remaining bandwidth for real-time traffic is used
virtual links are considered in the configuration problem. for other network traffic.
The proposed algorithm is based on branch-and-bound Since the AFDX network configuration becomes an
technique so that the simulation results show that it is faster important issue in avionics systems, we will investigate many
than the exhaustive search algorithm of the previous work. problems based on the results of this paper. For example,
We also provide an optimal solution of the problem for we will extend the problem into multiple AFDX switches or
the purpose of reducing the total bandwidth of configuration. discuss the routing issues through the networks.
International Journal of Aerospace Engineering 9

Conflict of Interests
The authors declare that there is no conflict of interests
regarding the publication of this paper.

Acknowledgments
This work was supported by Basic Science Research Program
through the National Research Foundation of Korea (NRF)
funded by the Ministry of Science, ICT & Future Planning
(no. NRF-2015R1A1A1A05001369), and the BK21 Plus Pro-
gram (Research Team for Software Platform on Unmanned
Aerial Vehicle, 21A20131600012) funded by the Ministry of
Education (MOE, Korea) and National Research Foundation
of Korea (NRF).

References
[1] R. L. Alena, J. P. Ossenfort, K. I. Laws, A. Goforth, and F.
Figueroa, “Communications for integrated modular avionics,”
in Proceedings of the IEEE Aerospace Conference, pp. 1–18, IEEE,
Big Sky, Mont, USA, March 2007.
[2] AFDX Tutorial, http://www.techsat.com/fileadmin/media/pdf/
infokiosk/TechSAT TUT-AFDX-EN.pdf.
[3] AFDX/ARNIC 664 Tutorial, http://www.cems.uwe.ac.uk/∼ngun-
ton/afdx detailed.pdf.
[4] A. Al Sheikh, O. Brun, M. Chéramy, and P.-E. Hladik, “Optimal
design of virtual links in AFDX networks,” Real-Time Systems,
vol. 49, no. 3, pp. 308–336, 2013.
[5] I. Saha and S. Roy, “A finite state modeling of AFDX frame
management using spin,” in Formal Methods: Applications and
Technology, vol. 4346 of Lecture Notes in Computer Science, pp.
227–243, Springer, Berlin, Germany, 2007.
[6] J. Täubrich and R. von Hanxleden, “Formal specification and
analysis of AFDX redundancy management algorithms,” in
Computer Safety, Reliability, and Security, vol. 4680 of Lecture
Notes in Computer Science, pp. 436–450, Springer, Berlin,
Germany, 2007.
[7] H. Charara, J.-L. Scharbarg, J. Ermont, and C. Fraboul, “Meth-
ods for bounding end-to-end delays on an AFDX network,”
in Proceedings of the 18th Euromicro Conference on Real-Time
Systems (ECRTS ’06), pp. 193–202, Dresden, Germany, July
2006.
[8] M. Tawk, X. Liu, L. Jian, G. Zhu, Y. Savaria, and F. Hu, “Optimal
scheduling and delay analysis for AFDX end systems,” SAE
Technical Paper 2011-01-2751, SAE International, 2011.
[9] J. J. Gutierrez, J. C. Palencia, and M. G. Harbour, “Response
time analysis in AFDX networks with sub-virtual links and
prioritized switches,” in XV Jornadas de Tiempo Real, Santander,
2012.
[10] D. Song, X. Zeng, L. Ding, and Q. Hu, “The design and
implementation of the AFDX network simulation system,”
in Proceedings of the International Conference on Multimedia
Technology (ICMT ’10), pp. 1–4, Ningbo, China, October 2010.
[11] D. An, H. W. Jeon, K. H. Kim, and K. I. Kim, “A feasible
configuration of AFDX networks for real-time flows in avionics
systems,” in Proceedings of the International Workshop on Real-
Time and Distributed Computing in Emerging Applications
(REACTION ’13), Vancouver, Canada, Decebmer 2013.
 International Journal of

Rotating
Machinery

International Journal of
The Scientific
Engineering Distributed
Journal of
Journal of

Hindawi Publishing Corporation

World Journal
Hindawi Publishing Corporation Hindawi Publishing Corporation
Sensors
Hindawi Publishing Corporation
Sensor Networks
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Journal of

Control Science
and Engineering

Advances in
Civil Engineering
Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Submit your manuscripts at

http://www.hindawi.com

Journal of
Journal of Electrical and Computer
Robotics
Hindawi Publishing Corporation
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

VLSI Design
Advances in
OptoElectronics
International Journal of

International Journal of
Modelling &
Simulation
Aerospace
Hindawi Publishing Corporation Volume 2014
Navigation and
Observation
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
in Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com
http://www.hindawi.com Volume 2014

International Journal of
International Journal of Antennas and Active and Passive Advances in
Chemical Engineering Propagation Electronic Components Shock and Vibration Acoustics and Vibration
Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014