Comparative performance evaluation of routing algorithm and topology size for wireless network-on-chip

Bulletin of Electrical Engineering and Informatics
Vol. 8, No. 4, December 2019, pp. 1239~1250
ISSN: 2302-9285, DOI: 10.11591/eei.v8i4.1588  1239
Journal homepage: http://beei.org/index.php/EEI
Comparative performance evaluation of routing algorithm and
topology size for wireless network-on-chip
Asrani Lit1
, M. S. Rusli2
, M. N. Marsono3
1
Department of Electrical and Electronic Engineering, Universiti Malaysia Sarawak, Malaysia
1,2,3
Division of Electronic and Computer Engineering, Universiti Teknologi Malaysia, Malaysia
Article Info ABSTRACT
Article history:
Received Mar 29, 2019
Revised Apr 10, 2019
Accepted Jun 8, 2019
Wireless Network-on-Chip or WiNoC is an alternative to traditional planar
on-chip networks. On-chip wireless links are utilized to reduce latency
between distant nodes due to its capability to communicate with far-away
node within a single hop. This paper analyzes the impact of various routing
schemes and the effect of WiNoC sizes on network traffic distributions
compared to conventional mesh NoC. Radio hubs (4×4) are evenly placed on
WiNoC to analyze global average delay, throughput, energy consumption
and wireless utilization. For validation, three various network sizes (8×8,
16×16 and 32×32) of mesh NoC and WiNoC architectures are simulated on
cycle-accurate Noxim simulator under numerous traffic load distributions.
Simulation results show that WiNoC architecture with the 16×16 network
size has better average speedup (∼1.2×) and improved network throughputs
by 6.36% in non-uniform transpose traffic distribution. As the trade-off,
WiNoC requires 63% higher energy consumption compared to the classical
wired NoC mesh.
Keywords:
Performance analysis
Routing algorithm
Traffic distribution
Wireless NoC
Copyright © 2019 Institute of Advanced Engineering and Science.
All rights reserved.
Corresponding Author:
Asrani Lit,
Department of Electrical and Electronic Engineering,
Universiti Malaysia Sarawak,
94300, Kota Samarahan, Sarawak, Malaysia.
Email: lasrani@unimas.my
1. INTRODUCTION
Technology scaling has allowed the integration of many-core design that range from hundreds to
thousands of processing cores on a single integrated circuit (IC) [1-3]. Nevertheless, the trend of growing
numbers of processing cores in Network-on-Chip (NoC) has caused higher latency and more power-hungry
system architecture affected by the long distance multiple-hops communication [4-6]. To date, Wireless
Network-on-Chip (WiNoC) is among the promising emerging technology to alleviate the aforementioned
issues due to its some unique factors such as energy efficiency, high bandwidth delivery and low latency
[4, 7-9]. Furthermore, these interconnects are also able to transmit data to distanced cores chip in single-hop
with minimal energy consumption [10-13].
On-chip wireless interconnects was proposed as the viable alternative for wired communication in
order to furnish an effective yet scalable WiNoC architecture [14]. There are numerous architectures of
WiNoC that have been proposed, which can be categorized as pure wireless, 2D mesh-based, multiple-tier,
small-world and irregular topology. Zhao et al. [15] proposed a multi-channel WiNoC called McWiNoC that
uses conventional NoC architecture as a basis, which is very adaptable to the transmission range of its radio
hubs. Meanwhile, a network-based processor array (NePA) that is a hybrid WiNoC that extends 2D mesh
NoC with a single bidirectional link between two neighbouring nodes was proposed [16]. Authors in [17]
have proposed an architecture known as WCube, a recursive wireless interconnects with a multi-tier structure
that includes both of the wireless backbone and wired edges. This architecture can cope with scaling

 ISSN: 2302-9285
Bulletin of Electr Eng and Inf, Vol. 8, No. 4, December 2019 : 1239 – 1250
1240
limitation of the demand number of on-chip cores. Instead of long wire insertion to improve NoC
performance [18], works in [19, 20] proposed a WiNoC architecture based on the small-world features.
Wireless links are used to create one-hop shortcuts. These alternatives have proven to enhance the network
performance in NoC. In [21], an irregular mesh-WiNoC topology that is established by wireless links has
been proposed to provide high efficient and low cost distributed minimal table based routing strategy.
Several research challenges in the design of WiNoC communication architectures have been raised
and in this research we are emphasis on investigating the impact of routing algorithm and topology size
[2, 4, 22, 23]. The routing protocol is important as in general it will influence the most critical network
metrics such as delay, throughput and power dissipation [24]. In addition, network topology designs equally
vital as the ability if the network to efficiently dissemminates packet depend on it underlying topology [25].
Therefore, the comprehensive understanding of efficient routing scheme and topology size in sentence traffic
patterns aid in the development of optimal network topology and impact on performance, power and design
cost in WiNoC architecture.
The main focus of this paper is the comparative performance evaluation of various routing schemes
and traffic patterns with different sizes of network topology in common WiNoC architecture. In this work we
analyzed the effect of increasing the network topology size and identify which architecture works best on
certain routing policy and traffic model with regard to delay, throughput and power consumption. The results
from this comparative evaluation enable and facilitate us to identify useful design trade-offs for optimal
development of wireless on-chip network based design. This paper is structured as follows. Section 2 reviews
the generic architecture of Wireless NoC. Section 3 discusses the popular routing algorithm for mesh network
topology. The experimental setup is highlighted in Section 4. Section 5 spotlights the simulation results and
discussion. Finally, the conclusion is given in Section 6 with recommendations for future work.
2. WIRELESS NOC ARCHITECTURE
The architecture formation in NoC is among the dominant factors that affect the overall performance
and its network cost [25, 26]. NoC topology is constructed by the physical layout, links between IP cores and
channels over the network. Therefore, the selection of topology will affect the hops count for a packet to
traverse and also the link lengths between source and destination IP cores. Hence, this explains how topology
impacts the network latency. Meanwhile, as data traversing process dissipates energy these hop counts
indirectly affects network energy consumption [27-29].
Most conventional NoC architectures are based on multiple-hops wired interconnections among
nodes. Unlike conventional NoC, WiNoC harnesses the wireless connectivity as a means of communication
among nodes with reduced network latency, particularly for the far-away nodes. Figure 1 illustrates the
generic example of 8×8 (64 nodes) mesh WiNoC architecture with 4×4 radio hubs that are distributed evenly
over the network. As depicted, WiNoC can be visualized as a two-level network. The first level is a classical
wired 8×8 mesh NoC topology, whereas the upper level is the wireless NoC constructed by a number of
radio-hubs. In this example, each radio-hub has a concentration of 4 that is shared by 4 routers of the first
level wired NoC network. NoC tiles are augmented with radio hub transceivers that allow single-hop
communication between far-away tiles that otherwise would require multiple-hops in the classical
wired network.
Figure 1. 8×8 WiNoC architecture with 4×4 radio hubs

Bulletin of Electr Eng and Inf ISSN: 2302-9285 
Comparative performance evaluation of routing algorithm and topology size for wireless network… (A. Lit)
1241
In WiNoC, the role of radio hubs is in charge of single-hop packet transfer between remote nodes
via wireless communication channel. The on-chip antenna and the transceiver are two essential modules in
radio hub. The features of both elements are discussed accordingly in Section 2.1 and 2.2.
2.1. On-chip antennas
To be practical for the WiNoC architecture the on-chip antenna must be wideband, sufficiently
small and highly efficient [30, 23]. Meanwhile, it has to furnish the best power gain at the minimal area
overhead. A metal zig-zag antenna has been demonstrated to have these criteria [31]. Furthermore, this
antenna also provides negligible effect of rotation on received signal strength, hence making it the most
appropriate candidate for WiNoC application [15]. Figure 2 shows the detail of the zig-zag antenna structure.
Figure 2. The zig-zag antenna [8, 32]
2.2. Wireless transceiver architecture
The low power design of wireless transceiver is the pivotal factor to ensure the desired performance
in the WiNoC system [33-36]. For that reason, low-power design considerations were taken into account for
both architecture and circuit levels of the transceiver [37-39]. As depicted in Figure 3, a generic WiNoC
OOK transceiver composed by a transmitter and a receiver that share the same antenna by means of a
RF-switch [31, 40, 41]. As shown in figure, a token controller is present to assure that the wireless channel is
not busy at the moment of a transmission [42]. If the channel is free incoming flit will be converted in a serial
fashion by mean of the serializer. The main task of transmitters consists in adapting the data incoming from
the electrical medium to the wireless medium by means of an antenna. In particular, a transmitter is
constituted by a serializer, which converts parallel streams of data (flits) in a serial fashion. An OOK
modulator converts data in higher frequency signal that will be delivered to the antenna via a power amplifier
(PA). The structure of the receiver is the opposite of the transmitter. Radio frequency signals will be
converted in a baseband stream of data with a demodulator. A deserializer converts a serial stream in to a flit.
Figure 3. OOK WiNoC transceiver [31]
3. ROUTING ALGORITHM FOR MESH TOPOLOGY
The routing algorithm used in the underlying on-chip communication network is essential aspect
that distinguishes various proposed NoC architectures [43]. In this section, we focus on well-known routing
algorithm for mesh topology. With respect to mesh topology, it is easy to accomplish a shortest path
deterministic routing by employing a simple variation of dimension order routing such as XY routing [44].
XY routing ensures deadlock and livelock freedom, but it provide no adaptiveness. This algorithm is a
table-less routing technique whereby each packet is routed first in X direction and after it reaches the same X
as the destination address, similarly moves along the perpendicular dimension.

 ISSN: 2302-9285
1242
A number of shortest path, deadlock-free, partially adaptive routing algorithms are based on
restricting certain turns are West-first, North-last and Negative-first [45]. As their names imply, West-first
requires that a packet is routed to the west direction first, if it is a productive direction, otherwise any shortest
path can be taken. Similarly, North-last requires that if north is productive direction, to be taken last, while
negative first requires that. All these schemes imply that certain turns are prohibited as shown in Figure 4. As
can be seen in Figure 4(a), XY routing, by restricting four turns becomes completely deterministic. The
partially adaptive routing algorithms restricts only two turns. West-first, by requiring that the west direction
is taken first if required, allows all possible shortest path when a packet has to travel eastward, clearly
favorings traffic toward the east direction. Likewise, North-last by demanding that traffic going toward the
north takes the north direction last, allow all possible shortest paths for traffic going toward the south but
again, only a single path for traffic going north. Clearly this algorithm favours traffic going from north to
south. Finally, Negative-first requires that in case the packet destination is toward any negative axis,
horizontal or vertical, along with any other direction, then the packet should be routed first toward that
negative axis direction and afterward toward to other direction. All these schemes imply that certain turns are
prohibited as shown in Figure 4(b)-(d).
The Odd-even model is one of the most popular partial adaptive wormhole routing algorithms in 2D
mesh on-chip interconnection network [46]. Unlike the turn model which prohibits certain turn in all
locations of the network, in the Odd-even model some turns are restricted only in even columns and some
other turns are prohibited in odd columns. Odd even rules can be described by these dual rules. First, east
turns cannot be taken in even columns as depicted in and cecond, north turns cannot be taken in odd column
as shown in Figure 4(e).
A routing algorithm called dynamically adaptive and deterministic (DyAD) combines the
advantages of both deterministic and adaptive routing schemes is presented in [47]. This approach is based
on the current network congestion since each router in the network continuously monitors its local network
load and makes decision based on this information. When the network is not congested, the DyAD router
works in a deterministic mode, thus enjoying the low routing latency enabled by deterministic routing. On the
contrary when the network becomes congested, the DyAD router switches back to the adaptive routing mode,
and thus avoids the congested links by exploiting other routing path.
Figure 4. Possible turns in turn model routing algorithm
(The solid lines indicate the allowable turns and the dash lines indicate unallowable turns)
4. EXPERIMENTAL SETUP
The evaluation between NoC and WiNoC architectures employed a cycle-accurate systemC based
simulator called Noxim [48]. Mesh-based topology is chosen because of its natural layout easily map to an
IC. In addition, due to its physical regularity this network is also scalable and adaptable with simple routing
algorithm. In order to identify how the architectures reacts in different conditions, the simulation are carried
out under various traffic scenarios namely hotspot, transpose and random. Simulations have been done for

1243
three network scales namely 8×8 (64 nodes), 16×16 (256 nodes), and 32×32 (1024 nodes). Table 1 shows the
simulation setup, while the traffic descriptions are shown in Table 2.
Table 1. Simulation setup
Parameter Descriptions
Network Sizes 8×8, 16×16, 32×32
Number of Radio Hub 4×4
Number of Channels 8
Simulation Time
Technology
100 000
65 nm
Clock Frequency 1 GHz
Switching Mechanism Wormhole
Radio Access Control Token Packet
Flit Size 32 bits
Routing Algorithm XY, west-first, north-last,
negative-first, odd-even,
DyAD
Wireless Data Rate 16 Gbps
Wireless Communication millimeter-wave
Table 2. Traffic patterns
Pattern Descriptions
Random Uniform distribution of traffic from source to
destination where each node sends packets to
others with the same probability.
Transpose Bit-permutation traffic adopting transpose matrix.
Shuffle Bit-permutation traffic from source to destination
with shifted order address.
For different cases of traffic distribution, the experiments are carried out based on 8×8, 16×16, and
32×32 network sizes. The large size up to 32×32 (1024 cores) WiNoC is chosen to investigate the impact of
radio hub allocation and placement for WiNoC architecture in handling large network. As previously
mentioned, the comparative analysis between WiNoC and mesh NoC are made in terms of global average
delay, network throughput and energy consumption
5. RESULTS AND ANALYSIS
5.1. Effects of routing algorithm
Routing is the procedure that is employed to deliver the flits onward to the sentence directions over
the network between from the source to its destination. The performance of NoC and WiNoC architecture
have been evaluated with the popular different routing scheme namely XY, west-first, north-last,
negative- first, odd-even and DyAD. The selection behind of these routing algorithm are because they are
based on wormhole switching mechanism that provide deadlock and livelock freedom in the two-dimensional
mesh topology [49].
To investigate the impact of routing algorithms in various topology sizes, three sizes (8×8, 16×16
and 32×32) of networks have been simulated under transpose traffic. The rationale behind the selection of
this non-uniform traffic distribution is because of its practicality for the real world application [50]. Figure 5
shows how the performance of the NoC and WiNoC architecture with respect to network load for 8×8
network size. As can be observed, the performances of both architectures are almost identical for all routing
schemes. However, from the context of routing algorithm, Odd-even is shown best algorithm since it can
cater higher network load (saturation point at PIR 0.02).

 ISSN: 2302-9285
1244
Figure 5. Comparative performance evaluation between NoC and WiNoC architecture under different routing
algorithm with 8×8 network size
As we increased the network size to 16×16 and 32×32, the performance of WiNoC is improved for
all routing algorithm in comparison with the conventional mesh-NoC. This is shown in Figure 6 and 7
respectively. For instance, as can be seen in Figure 6(e) WiNoC has more capability in handling more
network load in odd even routing algorithm which has the saturation point at 0.005 instead of 0.003 for NoC
mesh. The reason of this is due to in the odd-even routing scheme, some turns are prohibited only in even
column meanwhile some other turns are restricted in odd column. Hence, the degree of adaptiveness offered
by this scheme is higher than others scheme.

1245
5.2. Effects of Network Size
To investigate the effect of network size in both architectures, three various network sizes (64, 256
and 1024 nodes) have been simulated under numerous traffic distributions as described in Table 2. Odd-even
routing scheme were adopted due to its advantage in handling more network load as has been explained in
Section 5.1. This routing is utilized in the wired NoC layer until a wireless node with radio hub is found. Flits
are then sent to its destination through the one-hop wireless channel.
5.2.1. Impact on Packet Injection Rate
Saturation throughput can be defined as the network throughput at which packet injection rate (PIR)
begins to saturate. It is a common metric utilized to evaluate network performance [51, 52, 53]. At this point
of network saturation throughput, the system is not effective in handling the network loads anymore. When
PIR is set to 0.001, the rate of injection is 1 packet for every 1000 cycles. Since packets are set to 32 flits
size, hence it resulted in 0.0032 flits/node/cycle, which corresponds to 0.001 packet/node/cycle of
throughput. Figure 8 summarizes the comparison on the network latency of two simulated architectures
namely mesh-NoC and WiNoC under different traffic load distributions as we vary the network dimensions.
Figure 8. Global average delay comparisons for various network sizes under different traffic
load distributions

 ISSN: 2302-9285
1246
The latencies for both architectures are almost identical for all scenarios of traffic patterns and
network sizes at low network commitment. However, as the network load increases, the latency of network
begin to exponentially increase depending on traffic patterns and system sizes. In general, an architecture that
has higher PIR demonstrates a better system because of its ability to manage higher network loads. For
instance, in Figure 8(a), at transpose traffic WiNoC saturates at PIR 0.007 compared to NoC at PIR 0.009 has
reveals that NoC has better performance in 8×8 system size. In contrast, Figure 8(b) for transpose distribution
shows 16×16 WiNoC architecture is saturated at higher PIR, 0.006 compared to 0.004 for NoC which
indicates that WiNoC has more desirable performance. Overall, from the network size perspective WiNoCs
have disadvantages in 8×8 topology scale. In the contrary, as the topology scale become bigger (16×16 and
32×32) WiNoCs were shown to have more fitting performance due to the factor of radio hub that assist in
single-hop long distance wireless communication.
5.2.2. Impact on network throughput
Throughput is another important metric indicator of the performance and quality of a network
connection. It can be defined as how the network is able to process the requested packet injection rate and
can be represented as flits/cycle. A high ratio of unsuccessful packet delivery will lead to lower throughput
and degraded performance. Network throughput is affected by a number of factors such as network
congestion and packet loss. Hence, the higher the throughput in the network reflects to more
effective system.
Figure 9 shows the network throughput comparisons for various network sizes (8×8, 16×16 and
32×32) under different case of traffic scenarios. As illustrated in transpose traffic, bigger network WiNoC256
and WiNoC1024 have 6.36% and 1.82% respectively higher throughput in comparison with the conventional
NoC architecture. The reason for this is the ability of the bigger system architecture that can cope with higher
networks load as has been discussed in the impact of PIR in Section 5.2.1.
Figure 9. Network throughput comparisons for various network sizes under different traffic load distributions
5.2.3. Impact on energy consumption
The reduced energy consumption leads to a better power characteristic in the system architecture.
Figure 10 shows the energy comparisons between NoC and WiNoC for varied network sizes subject to
different type of traffic patterns. As can be observed, the energy consumption has an incremental trend as the
network increase in size. In addition, WiNoCs utilized more energy consumption for all cases networks sizes
and traffic distributions.
This is due to the wireless communication activities introduced by on-chip transceivers in WiNoC
composition. On the other hand, wireless radio hubs routers are more energy-hungry than conventional NoC
routers. From the context of network size, as in transpose traffic for example when network topology grows
bigger WiNoC consume lower energy consumption reduced from 87% (8×8), 63% (16×16) to 27% (32×32).
This revealed that WiNoC benefited in energy saving when the network size is larger. The factor of this
savings is due to the least wireless utilization of the radio hub in the bigger WiNoC as described in
Section 5.2.4.

1247
Figure 10. Energy consumption comparisons for various network sizes under different traffic
load distributions
5.2.4. Impact on wireless utilization
The considered traffic settings (uniform random, transpose and shuffle) allow the network to work
in different regions characterized by a different utilization of the wireless medium. In particular, wireless
utilization can defined as the ratio between the number of communications that use, totally or in part, the
wireless medium and the total number of communications. As shown, the percentages of wireless utilization
are inversely proportional with system size.
The high percentage (on average 86%) of wireless usage in 8×8 is because of the dense distributions
of 16 radio hubs in 64 nodes WiNoC. In the contrary, for biggest network (32×32) from this experiment
resulted in low percentage (on average 3%) of wireless usage due sparse distribution of the radio hubs.
Optimally, the 16×16 network architecture give the most suitable concentration that use on average 27%
wireless utilization. This is practical since the radio hubs were used for the purpose of long distance
communication. Hence, this justifies the reason why 256 nodes give the best results in terms of performance
as well as network throughput.
Figure 11. Percentage of wireless utilization for different size of WiNoC architecture under various
traffic distributions
6. CONCLUSION
The objective of this research is to investigate the impacts of different routing strategy and the
effects in varying the number of network sizes between the classical mesh NoC and WiNoC architecture.
From experimental results, can be concluded that WiNoC architecture performs its best among other
topology at 16×16 network size in nonuniform transpose traffic with better average speedup (~1.2×) and
improved network throughput by 6.36%. However, it has the trade-off on higher energy consumption in
comparison with conventional mesh NoC. For future work, we target to look into the implication of the
number radio hub placements on several specific WiNoC architectures such as iWise, WCube and
McWiNoC.

 ISSN: 2302-9285
1248
ACKNOWLEDGEMENTS
The first author would like to thank his employer Universiti Malaysia Sarawak (UNIMAS) and
Ministry of Education Malaysia for financing his PhD studies.
REFERENCES
[1] S. Borkar, “Thousand core chips: a technology perspective,” in Proceedings of the 44th annual Design Automation
Conference. ACM, 2007, pp. 746–749.
[2] A. B. Achballah, S. B. Othman, and S. B. Saoud, “Problems and challenges of emerging technology
networks-on-chip: A review,” Microprocessors and Microsystems, vol. 53, pp. 1–20, 2017.
[3] L. Benini and G. De Micheli, “Networks on chips: a new soc paradigm,” IEEE Computer, vol. 35, no. 1, pp. 70–78,
2002.
[4] A. Ganguly, M. M. Ahmed, R. Singh Narde, A. Vashist, M. S. Shamim, N. Mansoor, T. Shinde, S. Subramaniam,
S. Saxena, J. Venkataraman et al., “The advances, challenges and future possibili- ties of millimeter-wave
chip-to-chip interconnections for multi-chip systems,” Journal of Low Power Electronics and Applications, vol. 8,
no. 1, p. 5, 2018.
[5] D. Bertozzi, G. Dimitrakopoulos, J. Flich, and S. Sonntag, “The fast evolving landscape of on-chip
communication,” Design Automation for Embedded Systems, vol. 19, no. 1-2, pp. 59–76, 2015.
[6] M. Said, A. Shalaby, and F. Gebali, “Thermal-aware network-on-chips: Single-and cross-layered approaches,”
Future Generation Computer Systems, vol. 91, pp. 61–85, 2019.
[7] K. Chang, S. Deb, A. Ganguly, X. Yu, S. P. Sah, P. P. Pande, B. Belzer, and D. Heo, “Performance evaluation and
design trade-offs for wireless network-on-chip architectures,” ACM Journal on Emerging Technologies in
Computing Systems (JETC), vol. 8, no. 3, p. 23, 2012.
[8] S. Deb, K. Chang, A. Ganguly, X. Yu, C. Teuscher, P. Pande, D. Heo, and B. Belzer, “Design of an efficient noc
architecture using millimeter-wave wireless links,” in Proceedings of the 13th IEEE International Symposium on
Quality Electronic Design (ISQED 2012), California, USA, March 2012, pp. 165–172.
[9] D. Halperin, S. Kandula, J. Padhye, P. Bahl, and D. Wetherall, “Augmenting data center networks with
multi-gigabit wireless links,” in ACM SIGCOMM Computer Communication Review, vol. 41. ACM, 2011,
pp. 38–49.
[10] D. DiTomaso, A. Kodi, D. Matolak, S. Kaya, S. Laha, and W. Rayess, “A-winoc: Adaptive wireless
network-on-chip architecture for chip multiprocessors,” IEEE Transactions on Parallel and Distributed Systems,
vol. 26, no. 12, pp. 3289–3302, 2014.
[11] S. Abadal, E. Alarcón, A. Cabellos-Aparicio, M. Lemme, and M. Nemirovsky, “Graphene-enabled wireless
communication for massive multicore architectures,” IEEE Communications Magazine, vol. 51, no. 11,
pp. 137–143, 2013.
[12] S. Abadal, A. Mestres, J. Torrellas, E. Alarcón, and A. Cabellos-Aparicio, “Medium access control in wireless
network-on-chip: A context analysis,” IEEE Communications Magazine, vol. 56, no. 6, pp. 172–178, 2018.
[13] S. Abadal, J. Torrellas, E. Alarcón, and A. Cabellos-Aparicio, “Orthonoc: A broadcast-oriented dual- plane wireless
network-on-chip architecture,” IEEE Transactions on Parallel & Distributed Systems, no. 1, pp. 1–1, 2018.
[14] A. Karkar, T. Mak, K.-F. Tong, and A. Yakovlev, “A survey of emerging interconnects for on-chip efficient
multicast and broadcast in many-cores,” IEEE Circuits and Systems Magazine, vol. 16, no. 1, pp. 58–72, 2016.
[15] D. Zhao, Y. Wang, J. Li, and T. Kikkawa, “Design of multi-channel wireless noc to improve on-chip
communication capacity,” in Proceedings of the fifth ACM/IEEE International Symposium on Networks-on-Chip
(NoCS 2011), Pennsylvania, USA, May 2011, pp. 177–184.
[16] C. Wang, W.-H. Hu, and N. Bagherzadeh, “A wireless network-on-chip design for multicore plat- forms,” in
Proceedings of the 19th Euromicro International Conference on Parallel, Distributed and Network-Based
Processing (PDP 2011), Ayia Napa, Cyprus, February 2011, pp. 409–416.
[17] S.-B. Lee, S.-W. Tam, I. Pefkianakis, S. Lu, M. F. Chang, C. Guo, G. Reinman, C. Peng, M. Naik, L. Zhang et al.,
“A scalable micro wireless interconnect structure for cmps,” in Proceedings of the 15th Annual International
Conference on Mobile Computing and Networking (MobiCom 2009), Beijing, China, September 2009,
pp. 217–228.
[18] U. Y. Ogras and R. Marculescu, “"it’s a small world after all": Noc performance optimization via long-range link
insertion,” IEEE Transactions on Very Large Scale Integration (VLSI) Systems, vol. 14, no. 7, pp. 693–706, 2006.
[19] S. Deb, A. Ganguly, K. Chang, P. Pande, B. Beizer, and D. Heo, “Enhancing performance of network-on-chip
architectures with millimeter-wave wireless interconnects,” in Proceedings of the 21st IEEE International
Conference on Application-specific Systems Architectures and Processors (ASAP 2010), Rennes, France, July
2010, pp. 73–80.
[20] A. Ganguly, K. Chang, S. Deb, P. P. Pande, B. Belzer, and C. Teuscher, “Scalable hybrid wireless network-on-chip
architectures for multicore systems,” IEEE Transactions on Computers, vol. 60, no. 10, pp. 1485–1502, 2011.
[21] R. Wu, Y. Wang and D. Zhao, "A Low-Cost Deadlock-Free Design of Minimal-Table Rerouted XY-Routing for
Irregular Wireless NoCs," 2010 Fourth ACM/IEEE International Symposium on Networks-on-Chip, Grenoble,
2010, pp. 199-206.
[22] S. Deb, A. Ganguly, P. P. Pande, B. Belzer, and D. Heo, “Wireless noc as interconnection backbone for multicore
chips: Promises and challenges,” IEEE Journal on Emerging and Selected Topics in Circuits and Systems, vol. 2,
no. 2, pp. 228–239, 2012.

1249
[23] R. S. Narde, N. Mansoor, A. Ganguly, and J. Venkataraman, “On-chip antennas for inter-chip wireless
interconnections: Challenges and opportunities,” 2018.
[24] M. Palesi and M. Daneshtalab, Routing algorithms in networks-on-chip. Springer, 2014.
[25] R. Marculescu, U. Y. Ogras, L.-S. Peh, N. E. Jerger, and Y. Hoskote, “Outstanding research problems in noc
design: system, microarchitecture, and circuit perspectives,” IEEE Transactions on Computer-Aided Design of
Integrated Circuits and Systems, vol. 28, no. 1, pp. 3–21, 2009.
[26] U. Y. Ogras, J. Hu, and R. Marculescu, “Key research problems in noc design: a holistic perspective,” in
Proceedings of the 3rd IEEE/ACM/IFIP international conference on Hardware/software codesign and system
synthesis. ACM, 2005, pp. 69–74.
[27] M. S. Rusli, A. Lit, M. N. Marsono, and M. Palesi, “Adaptive packet relocator in wireless network-on-chip
(winoc),” in Asian Simulation Conference. Springer, 2017, pp. 719–735.
[28] A. Mineo, M. Palesi, G. Ascia, and V. Catania, “An adaptive transmitting power technique for energy efficient
mm-wave wireless nocs,” in Proceedings of IEEE Design, Automation and Test in Europe Conference and
Exhibition (DATE 2014), Dresden, Germany, March 2014, pp. 1–6.
[29] M. S. Rusli, A. Mineo, M. Palesi, G. Ascia, V. Catania, and M. Marsono, “A closed loop control based power
manager for winoc architectures,” in Proceedings of ACM International Workshop on Manycore Embedded
Systems (MES 2014), Minneapolis, USA, June 2014, pp. 60–63.
[30] S. H. Gade, S. S. Ram, and S. Deb, “Millimeter wave wireless interconnects in deep submicron chips: Challenges
and opportunities,” Integration, vol. 64, pp. 127–136, 2019.
[31] B. Floyd, C.-M. Hung et al., “Intra-chip wireless interconnect for clock distribution implemented with integrated
antennas, receivers, and transmitters,” IEEE Journal of Solid-State Circuits, vol. 37, no. 5, pp. 543–552, 2002.
[32] S. Deb, K. Chang, M. Cosic, A. Ganguly, P. P. Pande, D. Heo, and B. Belzer, “Cmos compatible many-core noc
architectures with multi-channel millimeter-wave wireless links,” in Proceedings of the Great Lakes Symposium on
VLSI (GLSVLSI 2012), Massachusetts, USA, May 2012, pp. 165–170.
[33] E. Masri et al., "Accurate Channel Models for Realistic Design Space Exploration of Future Wireless NoCs," 2018
Twelfth IEEE/ACM International Symposium on Networks-on-Chip (NOCS), Turin, 2018, pp. 1-8.
[34] H. K. Mondal, S. Kaushik, S. H. Gade and S. Deb, "Energy-Efficient Transceiver for Wireless NoC," 2017 30th
International Conference on VLSI Design and 2017 16th International Conference on Embedded Systems (VLSID),
Hyderabad, 2017, pp. 87-92.
[35] X. Timoneda, S. Abadal, A. Franques, D. Manessis, J. Zhou, J. Torrellas, E. Alarcón, and A. Cabellos-Aparicio,
“Engineer the channel and adapt to it: Enabling wireless intra-chip com-munication,” arXiv preprint
arXiv:1901.04291, 2018.
[36] D. Matolak, A. Kodi, S. Kaya, D. DiTomaso, S. Laha, and W. Rayess, “Wireless networks-on-chips: architecture,
wireless channel, and devices,” IEEE Wireless Communications, vol. 19, no. 5, 2012.
[37] V. Catania, A. Mineo, S. Monteleone, M. Palesi and D. Patti, "Improving the energy efficiency of wireless Network
on Chip architectures through online selective buffers and receivers shutdown," 2016 13th IEEE Annual Consumer
Communications & Networking Conference (CCNC), Las Vegas, NV, 2016, pp. 668-673.
[38] A. Mineo, M. Palesi, G. Ascia, and V. Catania, “Runtime tunable transmitting power technique in mm-wave winoc
architectures,” IEEE Transactions on Very Large Scale Integration (VLSI) Systems, vol. 24, no. 4,
pp. 1535–1545, 2016.
[39] V. Catania, A. Mineo, S. Monteleone, M. Palesi, and D. Patti, “Improving energy efficiency in wireless
network-on-chip architectures,” ACM Journal on Emerging Technologies in Computing Systems (JETC), vol. 14,
no. 1, p. 9, 2018.
[40] B. Floyd, L. Shi, Y. Taur, I. Lagnado et al., “A 23.8-ghz soi cmos tuned amplifier,” IEEE Transactions on
Microwave Theory and Techniques, vol. 50, no. 9, pp. 2193–2196, 2002.
[41] S. Abadal, A. Mestres, M. Nemirovsky, H. Lee, A. González, E. Alarcón, and A. Cabellos-Aparicio, “Scalability of
broadcast performance in wireless network-on-chip,” IEEE Transactions on Parallel and Distributed Systems, vol.
27, no. 12, pp. 3631–3645, 2016.
[42] M. Palesi, M. Collotta, A. Mineo, and V. Catania, “An efficient radio access control mechanism for wireless
network-on-chip architectures,” Journal of Low Power Electronics and Applications, vol. 5, no. 2,
pp. 38–56, 2015.
[43] T. Bjerregaard and S. Mahadevan, “A survey of research and practices of network-on-chip,” ACM Computing
Surveys (CSUR), vol. 38, no. 1, p. 1, 2006.
[44] K. Goossens, J. Dielissen, and A. Radulescu, “Æthereal network on chip: concepts, architectures, and
implementations,” IEEE Design & Test of Computers, vol. 22, no. 5, pp. 414–421, 2005.
[45] C. J. Glass and L. M. Ni, “The turn model for adaptive routing,” ACM SIGARCH Computer Architecture News, vol.
20, no. 2, pp. 278–287, 1992.
[46] G.-M. Chiu, “The odd-even turn model for adaptive routing,” IEEE Transactions on parallel and distributed
systems, vol. 11, no. 7, pp. 729–738, 2000.
[47] J. Hu and R. Marculescu, “Dyad: smart routing for networks-on-chip,” in Proceedings of the 41st annual Design
Automation Conference. ACM, 2004, pp. 260–263.
[48] V. Catania, A. Mineo, S. Monteleone, M. Palesi, and D. Patti, “Noxim: An open, extensible and cycle-accurate
network on chip simulator,” in Proceedings of the IEEE 26th International Conference on Application-specific
Systems, Architectures and Processors (ASAP 2015), Ontario, Canada, July 2015, pp. 162–163.
[49] P. Mohapatra, “Wormhole routing techniques for directly connected multicomputer systems,” ACM Computing
Surveys (CSUR), vol. 30, no. 3, pp. 374–410, 1998.

 ISSN: 2302-9285
1250
[50] K. Tatas, K. Siozios, D. Soudris, and A. Jantsch, Designing 2D and 3D network-on-chip architectures. Springer,
2014.
[51] J. Duato, S. Yalamanchili, and L. M. Ni, Interconnection networks: An engineering approach. Morgan Kaufmann,
Publishers An Imprint of Elsevier Science. 2003.
[52] J. L. Hennessy and D. A. Patterson, “Computer architecture: a quantitative approach”. Elsevier, 2011.
[53] W. J. Dally and B. P. Towles, “Principles and practices of interconnection networks”. Elsevier, 2004.
BIOGRAPHIES OF AUTHORS
Asrani Lit received his B. Eng. and M. Eng. degree in Microelectronics & Computer System
from Universiti Teknologi Malaysia. He is doctoral student in the Division of Electronic and
Computer Engineering at the same university. His research interests focus on Network-on-Chip
and on-chip interconnect architectures.
Mohd Shahrizal Rusli received his Ph.D in Electrical Engineering (2016), M. Eng. (Electronic
and Telecommunication) (2010) and B. Eng. in Computer (2006) degree from Universiti
Teknologi Malaysia (UTM), Johor. Between 2014 and 2015, he was attached to University of
California Irvine, USA and University of Catania, Italy under the supervision of Prof. Dr. Nader
Bagherzadeh and Assoc. Prof. Dr. Maurizio Palesi, respectively. He currently serves as senior
lecturer at the Division of Electronic and Computer Engineering, UTM where his research
interest and specialization are in the field of embedded processor system, deep learning
application in embedded system, low-power and energy management in network-on-chip and
wireless network-on-chip. He has published numerous journals, proceedings and a book chapter.
He received several recognitions internationally and nationally such as best paper and national
competition awards related to his research interest. He has been appointed as a reviewer for
several journals. He is also an active researcher in VLSI and Embedded Computing Architecture
Design (VeCAD) Research Group and has participated in several courses and seminars.
Muhammad Nadzir Marsono is an Associate Professor at Division of Electronic and Computer
Engineering, Universiti Teknologi Malaysia. His research interests include system level
integration, working in multi-processor System-on-Chip, Network-on-Chip, network
algorithmics, and network processor architectures.

Comparative performance evaluation of routing algorithm and topology size for wireless network-on-chip

More Related Content

What's hot (19)

Similar to Comparative performance evaluation of routing algorithm and topology size for wireless network-on-chip (20)

More from journalBEEI (20)

Recently uploaded (20)

Comparative performance evaluation of routing algorithm and topology size for wireless network-on-chip