FCA- OE: Fault aware and congestion aware routing algorithm based on Odd-Even algorithm for network on chip

Đăng ngày 4/2/2019 3:56:05 PM | Thể loại: | Lần tải: 0 | Lần xem: 2 | Page: 5 | FileSize: 0.28 M | File type: PDF
FCA- OE: Fault aware and congestion aware routing algorithm based on Odd-Even algorithm for network on chip. The Simulation results prove the FCA-OE routing algorithm can improve the average packet latency and the average packet loss rate (10%) in the presence of permanent faults compared with OE routing algorithm and it is capable to route the packets even with the faulty switches and links in network. Simulation results demonstrate that the FCA-OE routing algorithm provides less average energy consumption as compared to the Fault-Aware Dynamic Routing algorithm.
International Journal of Computer Networks and Communications Security
VOL. 3, NO. 3, MARCH 2015, 83–87
Available online at: www.ijcncs.org
E-ISSN 2308-9830 (Online) / ISSN 2410-0595 (Print)
FCA- OE: Fault-Aware and Congestion-Aware Routing Algorithm
Based on Odd-Even Algorithm for Network on Chip
Amin Mehranzadeh1, Mehdi Hoodgar2 and Faraz Forootan3
1, 2, 3 Department of computer, Dezful Branch, Islamic Azad University, Dezful, Iran
E-mail: 1mehranzadeh@gmail.com, 2hoodgar@gmail.com, 3forootan@gmail.com
ABSTRACT
Routing algorithms play a crucial role in design of the network on chip .Fault and congestion are two
situations that effect on the packets latency and energy consumption of NoC. This paper presents a fault and
a congestion aware routing algorithm called FCA-OE based on Odd-Even (OE) routing algorithm. The
simulation shows the strength of the FCA-OE by comparing it with odd-even routing algorithm under
different traffic patterns. The Simulation results prove the FCA-OE routing algorithm can improve the
average packet latency and the average packet loss rate (10%) in the presence of permanent faults compared
with OE routing algorithm and it is capable to route the packets even with the faulty switches and links in
network. Simulation results demonstrate that the FCA-OE routing algorithm provides less average energy
consumption as compared to the Fault-Aware Dynamic Routing algorithm.
Keywords: Network-on-Chip, Routing Algorithm, Adaptive, Fault Aware, Congestion Aware.
1
INTRODUCTION
The
network
on
chip
(NoC)
is
a
layered
algorithms the path selected only by the source and
architecture
[1]
has been
proposed
for
improve
destination
nodes.
But,
in
the
adaptive
routing
performance
and
scalability
of
system
on
chip
algorithms, the route taken depends on dynamic
(SoC). Figure 1 shows an abstract view of a NOC
network conditions with given a beginning and a
in this architecture. As shown, each tile consists of
destination
address.
The
deterministic
routing
a resource (show as R) and a switch or router (show
algorithms
because
of
their
simplicity
used
for
as S). In NoC each resource is connected to the
routers
design
and
provide
low
latency
when
switch by a network interface and each switch is
congestion of the network is low. However, the
connected to the four neighboring tiles via channels
deterministic routing algorithms are likely to suffer
and sending packets via a path consisting of a series
from throughput degradation when congestion of
of switches and channels [2, 3]. The functioning of
the
network
increases
and
cannot
dynamically
network on chip is dependable of effectiveness of
responds
to
network
situations.
The
adaptive
the routing algorithms. Routing algorithms choose
routing algorithms by using alternative paths avoid
the best path between source and destination nodes
congestion
and
improve
the
performance
of
and Communication and performance of the entire
network on chip. However, the adaptive routing
system are significantly affected by the routing
algorithms have a higher latency when congestion
algorithm [4]. The Routing algorithms, classified
of the network is low because of the need to the
into deterministic routing algorithms and adaptive
extra
logic
in
order
to
choose
a
better
path
routing
algorithms. The
deterministic
routing
compared to deterministic routing algorithms [5, 6].
84
A. Mehranzadeh et. al / International Journal of Computer Networks and Communications Security, 3 (3), March 2015
S
S
S
S
dynamic routing scheme for network on chip. The
R
R
R
R
proposed algorithm can locate the faulty links and
avoid them in order to prevent packet losses and
S
R
S
R
S
R
S
R
increase throughput of the network. In addition, the
presented algorithm has the advantage of spreading
the load over the whole network by using the stress
S
S
S
S
factors.
R
R
R
R
3
PROPOSED ROUTING ALGORITHM
S
R
S
R
S
R
S
R
This paper, present a new fault and congestion
aware routing algorithm based on Odd-Even
routing, namely fault and congestion aware OE
Fig. 1. The typical structure of a 4*4 NOC
(FCA-OE). Figure 2 shows the pseudo code of the
FCA-OE routing algorithm, in FCA-OE a packet
This paper presents a Fault-aware and congestion-
aware routing algorithm called FCA-OE. The
proposed routing algorithm can route packets when
the network has faulty links or switches. Simulation
results demonstrate the advantage of the FCA-OE
in terms of packet loss rate and average packet
latency as compared to the Odd-Even (OE) routing
algorithm when the network has permanent faults.
Because of lower average packet latency of the
FCA-OE routing algorithm the average packet loss
rate is low (less than 10%).
routing first according to odd-even routing
algorithm and then the best path selected by
comparing of the stress factor and the mask (fault
mask). The instant packet queue length of input
buffer of each router, stress factor, has been
employed to pass around the congestion
information [14] and the mask is employed in order
to prevent packet losses and increase the overall
throughput of the network. Our algorithm uses both
the stress factor and the acknowledgment signal to
reroute the packets around the links with permanent
faults and to distribute the network load in order to
2
RELATED WORKS
lessen the probability of congestion [15].
The idea of NOC is derived from distributed
computing and large scale networks. The routing
algorithms for NOC have some tight constraints on
memory and computing resources and should be
FCA-OE Routing Algorithm Source, Destination, SwitchAddress) {
Declare OutPutSet;
If (Current Switch and Destination are in same Col and Row) {
Send packet to local & exit;
} else {
If (Current Switch and Destination are in same Col) {
If (Destination Row > Current Switch Row) {
Add south to OutPutSet;
simple and have unique design considerations
besides low latency and high throughput [7]. A
routing algorithm which is based on Odd-Even
routing algorithm and used both deterministic and
adaptive routing together is proposed in [3, 8]. In
this scheme, when congestion of the network is low
the switches work in deterministic mode and work
in adaptive mode when congestion of the network
} else {
Add north OutPutSet;
}} else {
If (Destination Col > Current Switch Col) { //Eastbound Messages
If (Current Switch and Destination are in same Row) {
Add East to OutPutSet;
} else {
If ((Current Switch Col Is Odd) OR (Current Switch Col == Source Col)) {
If (Destination Row > Current Switch Row) {
Add south to OutPutSet;
} else {
Add north to OutPutSet;
}
}
If ((Destination Col Is Odd) OR ((Destination Col – Current Switch Col) ! =1)) {
Add EAST to OutPutSet;
}
increase. For tolerant of the faults, several
researches worked on different routing schemes. In
order to successful transmission of messages
through the network, several copies of the same
packet sent along the network links [9]. Similarly,
by sending the flood packets to the network as
probabilistic flooding algorithms, finally packets
}
} else { // westbound
Add West to OutPutSet;
If (! IsOdd (switchCol)) {
If (Destination Row > Current Switch Row) {
Add south to OutPutSet;
} else if (Destination and Current Switch is not in same Row) {
Add north to OutPutSet;
}
}
}
}
}
//Select a Dimension from OutPutSet to forward the Packet;
If (there is more than one link) {
reach to the destination [10]. The probabilistic
flooding algorithms increase congestion when the
network is heavily loaded. In [11], to tolerate the
permanent faults, different deterministic routing
algorithms have been proposed to avoid of the
faulty links. In [12], a fault tolerant routing scheme
was suggested to avoid of the faulty links. This
If (All Links are Mask) {
Output link is -1;
} else if ((Link1! = Mask) and (Link2==Mask)) {
Output link is Link1;
} else if ((Link1 == Mask) and (Link2! =Mask)) {
Output link is Link2;
} else if ((Link1! = Mask) and (Link2! =Mask)) {
if ( Stress of Link1 <= Stress of Link2){
Output link is Link1;
} else {
Output link is Link2;
}
}
}
paper,
introduce
a
fault
and
congestion
aware
Fig. 2. Pseudo code of FCA-OE routing algorithm
85
A. Mehranzadeh et. al / International Journal of Computer Networks and Communications Security, 3 (3), March 2015
4
EXPERIMENTAL RESULTS
Avg Packet Received Percent
A Java based simulator, developed to evaluate
120
the FCA-OE routing algorithm [16]. The simulation
100
is run with OE and FCA-OE routing algorithms and
data are collected after a warm-up period of 10000
cycles. The Size of the mesh network is set to be
6×6 tiles and all of input buffers have 5 flits size
and use first in first out strategy. The congestion
threshold set at 60% of the total buffer's capacity.
80
60
40
20
0
0
10 20 30 40 50 60 70 80 90 100
Fault Percent
OE
FCA-OE
The performance of each type of routing algorithm
is evaluated through average packet loss percent
and throughput curves. Throughput defines as
follows:
Fig. 4. Average packet received rate under uniform
traffic
Throughput = Total received flits
Number of nodes×TotalCycles
(1)
In order to assess the efficiency of proposed fault
and congestion aware dynamic routing algorithm,
we carried on a series of experiments to comparing
The performance of each type of routing
algorithm is evaluated through average packet loss
percent and throughput curves. Where Total
received flits refers to the number of whole flits that
arrive at their destination, Number of nodes is the
number of nodes, and Total cycles is the number of
clock cycles between the time of generation of the
first message and reception of the last message.
Figure 3, shows the throughput under uniform
traffic pattern. As shown in Figure 3, FCA-OE
routing algorithm performs better than OE under
different fault percent schemes. The fault percent is
the number of faulty switches in the network that
have one link with permanent fault. The packet
injection rate is the number of packets injected into
the network per cycle (fixed to 150
(packets/cycle)).
FCA-OE efficiency against the Fault-aware
Dynamic Routing algorithm (FADR) [15]. The
FADR routing algorithm in terms of functionality,
latency, and energy consumption in the presence
permanent faults can achieve less latency and
provides less energy consumption compared to
recently reported fault tolerant routing algorithms
such as the flooding algorithms [10]. The
performance of the proposed routing algorithm is
evaluated through latency-throughput curves [3, 8].
The packet latency is the duration from the time
when the first flit is created at the source node, to
the time when the last flit is delivered to the
destination node. A simulation is conducted to
evaluate the average packet latency for a given
packet injection rate. For each simulation, latencies
are collected after the first 5,000 cycles to allow the
network to stabilize and the packet latencies are
Network Throughput
averaged over 50,000 packets. The packets have a
fixed length of 5 flits and the buffer size of input
0.18
channels is 5 flits.
0.16
0.14
0.12
0.1
OE
0.08
FCA-OE
0.06
0.04
0.02
0
0
10 20 30 40 50 60 70 80 90 100
Fault Percent
Fig. 3. Network throughput for OE and FCA-OE routing
algorithms
Figure
4,
shows
the
average
packet
received
percent under uniform traffic patterns. As shown in
this figure, the packet loss rate decreases about 10
percent by FCA-OE.
Fig. 5. Average packet latency under uniform traffic
when fault percent=0
net Throughput
Avg Packet Received
Percent
HƯỚNG DẪN DOWNLOAD TÀI LIỆU

Bước 1:Tại trang tài liệu slideshare.vn bạn muốn tải, click vào nút Download màu xanh lá cây ở phía trên.
Bước 2: Tại liên kết tải về, bạn chọn liên kết để tải File về máy tính. Tại đây sẽ có lựa chọn tải File được lưu trên slideshare.vn
Bước 3: Một thông báo xuất hiện ở phía cuối trình duyệt, hỏi bạn muốn lưu . - Nếu click vào Save, file sẽ được lưu về máy (Quá trình tải file nhanh hay chậm phụ thuộc vào đường truyền internet, dung lượng file bạn muốn tải)
Có nhiều phần mềm hỗ trợ việc download file về máy tính với tốc độ tải file nhanh như: Internet Download Manager (IDM), Free Download Manager, ... Tùy vào sở thích của từng người mà người dùng chọn lựa phần mềm hỗ trợ download cho máy tính của mình  
2 lần xem

FCA- OE: Fault aware and congestion aware routing algorithm based on Odd-Even algorithm for network on chip. The Simulation results prove the FCA-OE routing algorithm can improve the average packet latency and the average packet loss rate (10%) in the presence of permanent faults compared with OE routing algorithm and it is capable to route the packets even with the faulty switches and links in network. Simulation results demonstrate that the FCA-OE routing algorithm provides less average energy consumption as compared to the Fault-Aware Dynamic Routing algorithm..

Nội dung

International Journal of Computer Networks and Communications Security VOL. 3, NO. 3, MARCH 2015, 83–87 Available online at: www.ijcncs.org E-ISSN 2308-9830 (Online) / ISSN 2410-0595 (Print) FCA- OE: Fault-Aware and Congestion-Aware Routing Algorithm Based on Odd-Even Algorithm for Network on Chip Amin Mehranzadeh1, Mehdi Hoodgar2 and Faraz Forootan3 1, 2, 3 Department of computer, Dezful Branch, Islamic Azad University, Dezful, Iran E-mail: 1mehranzadeh@gmail.com, 2hoodgar@gmail.com, 3forootan@gmail.com ABSTRACT Routing algorithms play a crucial role in design of the network on chip .Fault and congestion are two situations that effect on the packets latency and energy consumption of NoC. This paper presents a fault and a congestion aware routing algorithm called FCA-OE based on Odd-Even (OE) routing algorithm. The simulation shows the strength of the FCA-OE by comparing it with odd-even routing algorithm under different traffic patterns. The Simulation results prove the FCA-OE routing algorithm can improve the average packet latency and the average packet loss rate (10%) in the presence of permanent faults compared with OE routing algorithm and it is capable to route the packets even with the faulty switches and links in network. Simulation results demonstrate that the FCA-OE routing algorithm provides less average energy consumption as compared to the Fault-Aware Dynamic Routing algorithm. Keywords: Network-on-Chip, Routing Algorithm, Adaptive, Fault Aware, Congestion Aware. 1 INTRODUCTION The network on chip (NoC) is a layered architecture [1] has been proposed for improve performance and scalability of system on chip (SoC). Figure 1 shows an abstract view of a NOC in this architecture. As shown, each tile consists of a resource (show as R) and a switch or router (show as S). In NoC each resource is connected to the switch by a network interface and each switch is connected to the four neighboring tiles via channels and sending packets via a path consisting of a series of switches and channels [2, 3]. The functioning of network on chip is dependable of effectiveness of the routing algorithms. Routing algorithms choose the best path between source and destination nodes and Communication and performance of the entire system are significantly affected by the routing algorithm [4]. The Routing algorithms, classified into deterministic routing algorithms and adaptive routing algorithms. The deterministic routing algorithms the path selected only by the source and destination nodes. But, in the adaptive routing algorithms, the route taken depends on dynamic network conditions with given a beginning and a destination address. The deterministic routing algorithms because of their simplicity used for routers design and provide low latency when congestion of the network is low. However, the deterministic routing algorithms are likely to suffer from throughput degradation when congestion of the network increases and cannot dynamically responds to network situations. The adaptive routing algorithms by using alternative paths avoid congestion and improve the performance of network on chip. However, the adaptive routing algorithms have a higher latency when congestion of the network is low because of the need to the extra logic in order to choose a better path compared to deterministic routing algorithms [5, 6]. 84 A. Mehranzadeh et. al / International Journal of Computer Networks and Communications Security, 3 (3), March 2015 S S S S R R R R S S S S R R R R S S S S dynamic routing scheme for network on chip. The proposed algorithm can locate the faulty links and avoid them in order to prevent packet losses and increase throughput of the network. In addition, the presented algorithm has the advantage of spreading the load over the whole network by using the stress factors. R R R R 3 PROPOSED ROUTING ALGORITHM S S S S R R R R Fig. 1. The typical structure of a 4*4 NOC This paper presents a Fault-aware and congestion-aware routing algorithm called FCA-OE. The proposed routing algorithm can route packets when the network has faulty links or switches. Simulation results demonstrate the advantage of the FCA-OE in terms of packet loss rate and average packet latency as compared to the Odd-Even (OE) routing algorithm when the network has permanent faults. Because of lower average packet latency of the FCA-OE routing algorithm the average packet loss rate is low (less than 10%). 2 RELATED WORKS The idea of NOC is derived from distributed computing and large scale networks. The routing algorithms for NOC have some tight constraints on memory and computing resources and should be simple and have unique design considerations besides low latency and high throughput [7]. A routing algorithm which is based on Odd-Even routing algorithm and used both deterministic and adaptive routing together is proposed in [3, 8]. In this scheme, when congestion of the network is low the switches work in deterministic mode and work in adaptive mode when congestion of the network increase. For tolerant of the faults, several researches worked on different routing schemes. In order to successful transmission of messages through the network, several copies of the same packet sent along the network links [9]. Similarly, by sending the flood packets to the network as probabilistic flooding algorithms, finally packets reach to the destination [10]. The probabilistic flooding algorithms increase congestion when the network is heavily loaded. In [11], to tolerate the permanent faults, different deterministic routing algorithms have been proposed to avoid of the faulty links. In [12], a fault tolerant routing scheme was suggested to avoid of the faulty links. This paper, introduce a fault and congestion aware This paper, present a new fault and congestion aware routing algorithm based on Odd-Even routing, namely fault and congestion aware OE (FCA-OE). Figure 2 shows the pseudo code of the FCA-OE routing algorithm, in FCA-OE a packet routing first according to odd-even routing algorithm and then the best path selected by comparing of the stress factor and the mask (fault mask). The instant packet queue length of input buffer of each router, stress factor, has been employed to pass around the congestion information [14] and the mask is employed in order to prevent packet losses and increase the overall throughput of the network. Our algorithm uses both the stress factor and the acknowledgment signal to reroute the packets around the links with permanent faults and to distribute the network load in order to lessen the probability of congestion [15]. FCA-OE Routing Algorithm Source, Destination, SwitchAddress) { Declare OutPutSet; If (Current Switch and Destination are in same Col and Row) { Send packet to local & exit; } else { If (Current Switch and Destination are in same Col) { If (Destination Row > Current Switch Row) { Add south to OutPutSet; } else { Add north OutPutSet; }} else { If (Destination Col > Current Switch Col) { //Eastbound Messages If (Current Switch and Destination are in same Row) { Add East to OutPutSet; } else { If ((Current Switch Col Is Odd) OR (Current Switch Col == Source Col)) { If (Destination Row > Current Switch Row) { Add south to OutPutSet; } else { Add north to OutPutSet; } } If ((Destination Col Is Odd) OR ((Destination Col – Current Switch Col) ! =1)) { Add EAST to OutPutSet; } } } else { // westbound Add West to OutPutSet; If (! IsOdd (switchCol)) { If (Destination Row > Current Switch Row) { Add south to OutPutSet; } else if (Destination and Current Switch is not in same Row) { Add north to OutPutSet; } } } } } //Select a Dimension from OutPutSet to forward the Packet; If (there is more than one link) { If (All Links are Mask) { Output link is -1; } else if ((Link1! = Mask) and (Link2==Mask)) { Output link is Link1; } else if ((Link1 == Mask) and (Link2! =Mask)) { Output link is Link2; } else if ((Link1! = Mask) and (Link2! =Mask)) { if ( Stress of Link1 <= Stress of Link2){ Output link is Link1; } else { Output link is Link2; } } } Fig. 2. Pseudo code of FCA-OE routing algorithm 85 A. Mehranzadeh et. al / International Journal of Computer Networks and Communications Security, 3 (3), March 2015 4 EXPERIMENTAL RESULTS Avg Packet Received Percent A Java based simulator, developed to evaluate the FCA-OE routing algorithm [16]. The simulation is run with OE and FCA-OE routing algorithms and data are collected after a warm-up period of 10000 cycles. The Size of the mesh network is set to be 6×6 tiles and all of input buffers have 5 flits size and use first in first out strategy. The congestion threshold set at 60% of the total buffer's capacity. The performance of each type of routing algorithm is evaluated through average packet loss percent and throughput curves. Throughput defines as follows: Throughput = Total received flits (1) Number of nodes×TotalCycles The performance of each type of routing algorithm is evaluated through average packet loss percent and throughput curves. Where Total received flits refers to the number of whole flits that arrive at their destination, Number of nodes is the number of nodes, and Total cycles is the number of clock cycles between the time of generation of the first message and reception of the last message. Figure 3, shows the throughput under uniform traffic pattern. As shown in Figure 3, FCA-OE routing algorithm performs better than OE under different fault percent schemes. The fault percent is the number of faulty switches in the network that have one link with permanent fault. The packet injection rate is the number of packets injected into the network per cycle (fixed to 150 (packets/cycle)). Network Throughput 0.18 0.16 0.14 0.12 120 100 80 60 FCA-OE 40 20 0 0 10 20 30 40 50 60 70 80 90 100 Fault Percent Fig. 4. Average packet received rate under uniform traffic In order to assess the efficiency of proposed fault and congestion aware dynamic routing algorithm, we carried on a series of experiments to comparing FCA-OE efficiency against the Fault-aware Dynamic Routing algorithm (FADR) [15]. The FADR routing algorithm in terms of functionality, latency, and energy consumption in the presence permanent faults can achieve less latency and provides less energy consumption compared to recently reported fault tolerant routing algorithms such as the flooding algorithms [10]. The performance of the proposed routing algorithm is evaluated through latency-throughput curves [3, 8]. The packet latency is the duration from the time when the first flit is created at the source node, to the time when the last flit is delivered to the destination node. A simulation is conducted to evaluate the average packet latency for a given packet injection rate. For each simulation, latencies are collected after the first 5,000 cycles to allow the network to stabilize and the packet latencies are averaged over 50,000 packets. The packets have a fixed length of 5 flits and the buffer size of input channels is 5 flits. 0.1 OE 0.08 FCA-OE 0.06 0.04 0.02 0 0 10 20 30 40 50 60 70 80 90 100 Fault Percent Fig. 3. Network throughput for OE and FCA-OE routing algorithms Figure 4, shows the average packet received percent under uniform traffic patterns. As shown in this figure, the packet loss rate decreases about 10 percent by FCA-OE. Fig. 5. Average packet latency under uniform traffic when fault percent=0 86 A. Mehranzadeh et. al / International Journal of Computer Networks and Communications Security, 3 (3), March 2015 Fig. 5 (b). Average packet latency under uniform traffic when fault percent=10 As shown in Figure 5 (a) and Figure 5 (b), the strength of the FCA-OE routing algorithm is confirmed throughout the experiments by the fact that it achieves shorter average packet latency compared to FADR routing algorithm. The computation of average energy consumption included the number of hops that the flits travel to reach the destination as well as the number of extra flit copies sent through the network because of unacknowledged flits. As shown in Figure 6, in terms of average energy consumption, the proposed routing algorithm has a consistent behavior in the presence of few faults, which is 1.05× on average less energy consumption compared to the FADR routing algorithm. 300000 250000 200000 150000 100000 50000 0 0 10 30 Fault Percent FCA-OE FADR Fig. 6. Average energy consumption for transmitting packets (energy/packet/hop) 5 CONCLUSION & FUTURE WORK This paper presented a new approach for fault aware and congestion aware routing scheme on the network on chip, namely FCA-OE routing algorithm. The presented routing algorithm distributing the load over the whole network by considering the stress factors. This paper evaluated the result of permanent faults, and packet injection rates on the functioning of the FCA-OE routing algorithm. Simulation results proved the effectiveness of FCA-OE routing algorithm in terms of packet loss percent and latency in the presence of permanent faults compared to odd-even (OE) routing algorithm and proved our algorithm can achieve an average 10 percent less latency and average 10 percent less packet loss rate. The FCA-OE routing algorithm in terms of average energy consumption has less energy consumption as compared to the fault aware Dynamic Routing algorithm (FADR) [15] and recently reported fault tolerant routing algorithms [10]. 7 REFERENCES [1] L. Benini and G.D .Micheli, "Networks on chips: a new SOC paradigm", IEEE Computer, Jan 2002, 35:70–78. [2] J. Henkel, W. Wolf and S. Chakradhar, "On-chip networks: A scalable, communication-centric embedded system design paradigm", VLSI Design, India, 2004, pp. 845-851. [3] J. Hu and R. Marculescu, "DyAD - Smart routing for networks-on-chip", DAC, USA, 2004, pp. 260-263. [4] T. Schonwald, J. Zimmermann, O. Bringmann, "Fully Adaptive Fault-Tolerant Routing Algorithm for Network-on-chip Architectures", 10th Euromicro Conference on Digital System Design Architectures Methods and Tools (DSD), 2007. [5] W. J. Dally and B. Towles, "Route packets, not wires: On-chip interconnection networks", DAC, USA, 2001, pp. 684-689. [6] Al. Hashimi, B.M. and Schmitz, "Improving Routing Efficiency for Network-on-Chip through Contention-Aware Input Selection", 11th Conference (ASP-DAC), Japan, 2006. [7] T. T. Ye, L. Benini, and G. De Micheli, "Packetization and routing analysis of on-chip multiprocessor networks", Journal of Systems Architecture, 2004, vol. 50, pp. 81-104. [8] G.M. Chiu, "The odd-even turn model for adaptive routing", IEEE Transactions on Parallel and Distributed Systems, 2000, Vol. 11, pp. 729-38. [9] S. Manolache, P. Eles, and Z. Peng, "Fault and energy-aware communication mapping with guaranteed latency for applications implemented on NOC", in Proceedings of IEEE/ACM Design Automation Conference, DAC, 2005, pp. 266–269. [10]M. Pirretti, G. Link, R. Brooks, N. Vijaykrishnan, M. Kandemir, and M. Irwin, 87 A. Mehranzadeh et. al / International Journal of Computer Networks and Communications Security, 3 (3), March 2015 "Fault tolerant algorithms for network-on-chip interconnect", in Proceedings of IEEE Computer Society Annual Symposium on VLSI, 2004, pp. 46–51. [11]D. Greenfield, A. Banerjee, J.G. Lee, and S. Moore, "Implications of rent’s rule for NOC design and its fault-tolerance", in Proceedings of International Symposium on Networks-on-Chip, 2007, pp. 283–294. [12]M. Ali, M. Welzl, and S. Hessler, "A fault tolerant mechanism for handling permanent and transient failures in a network on chip", in Proceedings of International Conference on Information Technology, ITNG, 2007, pp. 1027–1032. [13]L.M. Ni and P.K. Mckinley, "A survey of wormhole routing techniques in direct networks", IEEE Int, Conference on Computer, Feb. 1993, 26:62-76. [14]M. Li, Q.A. Zeng, and W.B. Jone, "DyXY - A Proximity Congestion-Aware Deadlock-Free Dynamic Routing Method for Network on Chip", in Proceedings of IEEE/ACM Design Automation Conference, DAC, 2006, pp. 849– 852. [15]A. Hosseini, T. Ragheb, Y. Massoud, "A Fault-Aware Dynamic Routing Algorithm for on-Chip-Networks", IEEE, 2008. [16]H. Hossain, M. Ahmed, A. Al-Nayeem, "GPNOCSim-A General Purpose Simulator for Network-on-chip", Dhaka, Bangladesh, ICICT, 2007.

Tài liệu liên quan