FPGA IMPLEMENTATION OF PRIORITYARBITER BASED ROUTER DESIGN FOR NOC SYSTEMS

http://www.iaeme.com/IJARET/index.asp 509 editor@iaeme.com
International Journal of Advanced Research in Engineering and Technology (IJARET)
Volume 10, Issue 2, March - April 2019, pp. 509-516, Article ID: IJARET_10_02_050
Available online at http://www.iaeme.com/IJARET/issues.asp?JType=IJARET&VType=10&IType=02
ISSN Print: 0976-6480 and ISSN Online: 0976-6499
© IAEME Publication
FPGA IMPLEMENTATION OF PRIORITY-
ARBITER BASED ROUTER DESIGN FOR NOC
SYSTEMS
Nirmala M
M.Tech Student,
Department of ECE, Bangalore Institute of Technology, VV Puram Bengaluru, India
Shylaja V
Assistant Professor
Department of ECE, Bangalore Institute of Technology, VV Puram Bengaluru, India
ABSTRACT
An efficient Priority-Arbiter based Router is designed along with 2X2 and 3X3 mesh
topology based NOC architecture are designed. The Priority –Arbiter based Router
design includes Input registers, Priority arbiter, and XY- Routing algorithm. The
Priority-Arbiter based Router and NOC 2X2 and 3X3 Router designs are synthesized
and implemented using Xilinx ISE Tool and simulated using Modelsim6.5f. The
implementation is done by Artix-7 FPGA device, and the physically debugging of the
NOC 2X2 Router design is verified using Chipscope pro tool. The performance results
are analyzed in terms of the Area (Slices, LUT’s), Timing period, and Maximum
operating frequency. The comparison of the Priority-Arbiter based Router is made
concerning previous similar architecture with improvements.
Keywords—NoC, SoC, Router, FPGA
Cite this Article: Nirmala M and Shylaja V, Fpga Implementation of Priority-Arbiter
Based Router Design for Noc Systems, International Journal of Advanced Research in
Engineering and Technology, 10(2), 2019, pp 509-516.
http://www.iaeme.com/IJARET/issues.asp?JType=IJARET&VType=10&IType=2
1. INTRODUCTION
The increased innovative ideas in the today technology have offered integration of many
platforms in one platform. The semiconductor sectors of today's genre are offering efficient and
powerful hardware complex structures. Due to the use of these structures the power usage will
be optimized. With the increased competition in the complex hardware design, many of the
researchers are working towards advancement. Similar way the software usage can be improved
with the Computer-aided design (CAD), and it can be developed. A CAD tool has simulation,
modeling, implementation and synthesis tools.

Nirmala M and Shylaja V
The system on chip (SoC) has different processing elements (PE) like intellectual properties
(IP), processors, etc., input and output modules for communication and memory blocks. The
present technology is very much interested in the on-chip elements.
The SoCs composed of many IP blocks, physical links, and routers. The IP block is used to
represent Digital signal processing, MPEG decoders, memory modules, etc. The Links and
routers of the SoC are used to build the communication for IP blocks or NoC organization. The
routers of the NoC will be connected to the IP blocks using input and output ports. The ports
required to build the connection is dependent on network topology.
The every NoC designs exhibit some features of performance factors like throughput,
latency, power usage, area, and hardware complexity. The above factors depend on each other,
which cannot be used individually. The performance factors can be split as cost and
performance parameters. The performance parameters of the NoC are considered as latency and
throughput. Generally, the latency considered as the time took to forward the packets to the
destination from the source, While the throughput is considered as the part of a packet which is
forwarded to the destination in particular time from the source. In any desired applications of
the NoC, we need low latency with high throughput. The low latency with the high throughput
can be obtained by proper utilization of the channel, implementation of the proper routing
algorithm, providing the livelock, deadlock & stardom freedom and by proper switching
technique implementation. In this sense there exist cost parameters like hardware complexity,
power usage, and area usage. The NoC design for the battery depending applications like mobile
devices suffer from power usability. The area is the primary factor in buffer size enhancement
from the single router. The buffer size will affect the router area and power usage. These buffers
size depends on the type of switching technology adaption and implemented routing algorithm.
The section 2 discuss about the existing works of Router and NoC and Problems finding.
The section 3 explains about the proposed Router with descriptions. The results and analysis of
the work is elaborated in section 4. Finally concludes the overall system with improvements
with future work in section 5.
2. RELATED WORKS
The review of the existing work on RISC Processor architectures and ALU Units are described
in below session.
The study of Bhanwala et al. [1] has focused on the design of reconfigurable Network on
chip (NOC) application, based on Field propagation array (FPGA). The design of this complete
application is done by the using of Verilog hardware description language. The presented
design application is worked in four channel east, west, north, and south and also work for a
crossbar switch, the presented all channels are work of first in first out multiplexers and buffers.
The work of Kapre et al. [2] have concentrated on the making of low cost but an active and
high-performance network-on-chip system based on Field propagation array. The quality of
FPGA is buffer less, smaller magnitude and run faster-operating frequency that's why it is used
with Network-on-chip system. The study of Daf et al. [3] has discussed the evaluation and
performance of Loopback Virtual channel router, in which also focused on the working process
of a router and discussed the router was helping component during working. The work of
Kashwan et al. [4] has concentrated on performance analysis of a new optimized Network-on-
chip router, in which using of multiprocessor network on a chip is a proper solution for
increasing a higher performance designed with VLSI. The work of Nasir et al. [5] has
concentrated on the problem of communication between sources to destination, in which
different types of issues come like area, delay, power consumption. For solving this problem
presented an Elastic buffering (EB) technique in virtual channelling in an in NoC. In which
pipeline flip-flop system is us as a storage location to store data.

Fpga Implementation of Priority-Arbiter Based Router Design for Noc Systems
The work of Aslam et al. [7] has concentrated on the development of a routing protocol
technique for efficient Network on-chip transmission, in here introduce a Junction-based
routing protocol about this protocol it is a suitable technique for log Network on chip platforms.
The study of Kumar et al. [8] have focused on the design of area-efficient router framework
based on FPGA for Network on chip, Once the completing of the design of channels, the
crossbar is also designed and after the completing the process of design of crossbar and all four
channels both designed are integrated with the making of the framework of a router. The work
of Lu et al. [9] have concentrated on the problem of delay in packet transmission and cost in
routing, for solving this issue presented a Low-Latency NoC Router framework. The study of
Monemi et al. [10] has focused on the problem of a design of less time taking network on the
chip with a low area overhead technique. For solving this problem presented a two-clock-cycle
latency network on chip micro-framework.
The work of Boutalbi et al. [11] has presented a trustable routing technique for the detection
of the online issue in dynamic Network on a chip. The presented technique can fully detection
and localization of inner and outer data packet during transmission and detects the error between
permanent and temporary error. The study of Wang et al. [12] have focused on the condition of
congestion and hot spot in traditional 3D Network on a chip, and this network is not an element
of multicast. For solving this hot spot problem in 3D network on chip presented a Multicast
Rotational Routing Arithmetic scheme (ARRAs). The work of Salah et al. [13] has concentrated
on the design of a scalable routing protocol for packet-switched in the network on chip in the
environment of System-on-chip. For the best packet delivery and solving the interconnection
issue presented a circuit-switched NoCs to a positive performance and presented quality of
service for a system on chip. Abdelrasoul et al. [14] have introduced a re-evaluate framework
named Round Robin Arbiter (RRA) based on FPGA to cover this gap. Abdelrasoul et al. [14]
have introduced a re-evaluate framework named Round Robin Arbiter (RRA) based on FPGA
to cover this gap. The work of Liu et al. [16] have discussed custom-silicon network-on-chip
into an FPGA; it's helpful on the communication bandwidth. For the efficient communication
presented a mixing Hard Network-on chip technique to focus on the packet switched routing
protocol and supported pre-schedule and packet switched communication. The study of
Siddhartha et al. [17] have presented a Hoplite-Q, Field propagation Gate Array-based Network
on chip routing technique, the presented technique is priority-aware friendly with FPGA its
capable to exploiting application priority during routing communication under a workload.
Problem statement: Most of the Existing NOC Router designs are software based and not
compactable to the real-time environments. In that, very few are hardware-based approaches and lacks
hardware complexities and performance metrics in real time. Most of the NOC Router designs are ASIC
Based approaches and limited to perform only a few parameters. Very few NOC Router verification is
done on real-time environments and lacks optimizations. To verify the hardware designs with real-time
data, it is difficult without using the simulation testbench. The Unit under Test (UUT) is the design
module is incorporated in the testbench to verify the designs. But it lacks with controllability in real
time.
Proposed Solution: The proposed Priority-arbiter based NOC Router overcomes comes to the
problems and improves the NOC performance. The NOC Router Verification is done quickly by using
intellectual property (IP) cores. In that, virtual input-output core (VIO) is used which improves the
controllability of the NOC in real-time verification.
3. PROPOSED ROUTER
The single router Design which includes Input Registers, Arbiter design, Packet formation, and
Routing Algorithm is shown in figure 1. The input register is having five port 16-bit inputs like
Packet input (local), east, west, south, and north inputs, these inputs are stored in register

temporarily. Once grants are ready in arbiter and temporary port data signals are stored in the
Priority Encoder.
The priority-based arbiter receives the requests from the five port inputs, based on the
request, the arbiter checks the all 5 possible states, which state is free to allocate the data based
on priority.
The FSM state machine is designed to check the grants based on the requests. Arbiter
generates 5- grants which are acts as a 5-bit select line to the priority encoder. Based on the 5-
bit select line, the priority encoder will works.
 If the select line is 00001, then packet input is selected.
 If the select line is 00010, then east input is selected.
 If the select line is 00100, then west input is selected.
 If the select line is 01000, then north input is selected.
 If the select line is 10000, then south input is selected.
Based on the priority encoder output, the frame the packet. Then apply the XY routing
algorithm to generate the five port output signals to reach the destination.
Input Register
Priority based Arbiter
Priority Encoder
And Packet
formation
XY Routing
algorithm
Packet Input
East Input
West
input
South
Input
North
Input
Packet output
East output
West output
South output
North output
Packet
GrantsInput
Requests
16-bit
16-bit
16-bit
16-bit
16-bit
16-bit
16-bit
16-bit
16-bit
16-bit
16-bit
16-bit
16-bit
16-bit
16-bit
5-bit
5-bit
16-bit
Figure 1 Single-Router Architecture.
The arbiter is used to select the request based on the priority and generates the grants. In
figure 2, the arbiter FSM is represented, and it is having three requests (req0, req1, and req2)
and three grants (gnt0, gnt1, and gnt2). In FSM, 4 states are there, and it acts as a present and
next state. The FSM works based on the binary encoding scheme for states. In design, States
are initialized to IDLE = 00, GRANT-0= 01, GRANT-1= 10, and GRANT-2= 11. The arbiter
FSM design, first finds the next state and then present state followed by output state.
The next state logic is same as combination logic. In four states, Based on the three input
requests (req0, req1, and req2), it generates the next state. If the state is IDLE, if the req0=1,
then next state is GRANT-0 else, if the req1=1, then next state is GRANT-1 else, if the req2=1,
then next state is GRANT-2.
If the state is GRANT-0 is the input state, if the req0=1 then immediately, the next state is
GRANT1. Otherwise, if the requests come from the req1 or req2, it will result in the priority

and stores those request temporarily. Once the next state releases the present requests, and it
gives priority to the stored temporary requests.
Similarly, If the state is GRANT-1 is the input state, if the req1=1 then immediately, the
next state is GRANT2. Otherwise, if the requests come from the req0 or req2, it will result in
the priority and stores those request temporarily. Once the next state releases the present
requests, and it gives priority to the stored temporary requests.
Similarly, If the state is GRANT-2 is the input state, if the req2=1 then immediately, the
next state is IDLE or GRANT0. Otherwise, if the requests come from the req0 or req1, it will
result in the priority and stores those request temporarily. Once the next state releases the
present requests, and it gives priority to the stored temporary requests.
IDLE
GRANT-0
GRANT-1
GRANT-2
Req0=1
Req1=1
Req2=1
Req1=1
& Req0 =0
Req2= 0
Req2=1
& Req1 =
0
Req1=0
Req0=0
Figure 2 Arbiter FSM Design
In the present state, IDLE is the default state. Based on the next state logic, the present state
is generated on every successive clock cycles. In output state, based on the present state inputs,
The 4 states generate the grants.
The packet formation is performed based on the priority encoded output data signals. It is
having mainly 8-bit packet data (priority encoded data), 3-bit is reversed for future purpose, 2-
bit Destination-X address, and 2-bit destination-Y address and 1-bit request.
The XY Routing algorithm is represented Based on the Current and destination address, the
framed packet will move X-direction or Y-direction as per algorithm.
 If the destination X (desX) is equal to Current X address and destination Y (desY)
is equal to Current Y address, then the local output will be generated.
 In that, if the destination Y (desY) is higher than the Current Y address, then the
south output will be generated. Otherwise, if the destination Y (desY) is less than
the Current Y address, then north output will be generated.
 Parallelly, if the destination X (desX) is greater than the Current X address, then the
east output will be generated. Otherwise, if the destination X (desX) is less than the
Current Y address, then the west output will be generated.
The NOC 2X2 router mainly consists of the four individual routers like R1, R2, R3, and
R4. Each router has 4-bit XY current address, for Router R1- 00_00, Router R2- 00_01, Router
R3- 01_00, and Router R4- 01_01.Based on the XY routing algorithm and destination address
the routing operation is performed in the NOC 2X2. For NOC 2X2 Router, for communicating

from one router to another router, linking wires are used. The router R1 receives the data
through west input (WI) of R3 and North input (NI) from R2 and generates east output (EO)
and south output (SO). Similarly, it is applicable for R2, R3, and R4. For NOC 3X3 Router, it
has nine routers R1-R9 which is similar to the NOC 2x2 router. The current -XY address will
be changed, and it consists of nine input and nine output ports.
4. RESULTS AND ANALYSIS
The proposed Router results are described detail in the below section. The Complete Router
with NoC is designed using Verilog HDL over Xilinx ISE Platform and simulated on Modelsim
simulator and Hardware prototyped on low cost Artix-7 FPGA.
The Single router Simulation Results are a representation shown in figure 3. Once clock
(clk) is activated with low reset, set the router inputs Local (packet_in), east (pine), west (pinw),
south (pins), and north (pinn). After performing the Routing operations based on current and a
destination address using XY algorithm, The 8-bit output of data_out (local output and 16-bits
east (poe), west (pow), south (pos), and north (pon) outputs are generated.
Figure 3 Single router Simulation Results
The proposed Single router is compared with similar previous router [18] concerning
resource utilization on same selected Spartan 3E -500 Device FPGA. The improvements in slice
Registers around 25 %, slice-FF's around 18%, 4-input LUT's is 11%, and Global Clock is
around 17%.
Table 1 Comparison of Single router with previous [18]
Logic
Utilizatio
n
Avail
able
Propose
d Router
Utiliz
ation
Router
[18]
Utiliz
ation
Slices 4656 862 18% 2024 43%
Slice Flip
Flops
9312 642 6% 2133 22%
4 input
LUTs
9312 1264 13% 2231 24%
bonded
IOBs
232 98 42% 74 32%
GCLKs 24 1 4% 5 21%
Overall the Single router is efficient and consumes less resource utilization than previous
architecture.

5. CONCLUSION
The cost-effective and straightforward Single router along with 2x2 NOC and 3X3 NOC
Routers are designed using Verilog-HDL and implemented on Artix-7 FPGA Platform. The
Single router mainly contains Input registers, Arbiter, and XY-Routing algorithm. The
Hardware architecture of 2x2 NOC using ICON IP Core and VIO IP cores are designed. The
simulation results of the NOC Routers is seen using Modelsim 6.5f. The NOC Routers are
implemented on Artix-7 FPGA with physically debugging using Chipscope pro tool. The NOC
Routers are synthesized and implemented using Xilinx ISE Tool. The Single Router is
compared with previous similar architecture on the same FPGA devices with an Improvement
in area overhead of slice Registers around 25 %, slice-FF's 18%, 4-input LUT's is 11% than the
previous single router.
In Future, these NoC routers are communicated with network interfaces and processing
elements for Systems on chip designs to reduce the chip complexity.
REFERENCES
[1] Bhanwala, Amit, Mayank Kumar, and Yogendera Kumar. "FPGA based design of low power
reconfigurable router for Network on Chip (NoC)." In Computing, Communication &
Automation (ICCCA), 2015 International Conference on, pp. 1320-1326. IEEE, 2015.
[2] Kapre, Nachiket, and Tushar Krishna. "FastTrack: Leveraging Heterogeneous FPGA Wires to
Design Low-cost High-performance Soft NoCs." In 2018 ACM/IEEE 45th Annual International
Symposium on Computer Architecture (ISCA), pp. 739-751. IEEE, 2018.
[3] Daf, Mamta P., and Bharti B. Saynkar. "Performance and Evaluation of Loopback Virtual
Channel Router with Heterogeneous Router for On-Chip Network." In Communication Systems
and Network Technologies (CSNT), 2014 Fourth International Conference on, pp. 1065-1069.
IEEE, 2014.
[4] Kashwan, K. R., and G. Selvaraj. "Implementation and performance analyses of a novel
optimized NoC router." In Convergence of Technology (I2CT), 2014 International Conference
for, pp. 1-6. IEEE, 2014.
[5] Gupta, Priyank, Ali Akoglu, Kathleen Melde, and Janet Roveda. "FPGA based single cycle, a
reconfigurable router for NoC applications." In Circuits and Systems (ISCAS), 2013 IEEE
International Symposium on, pp. 2428-2431. IEEE, 2013.
[6] Nazir, Liyaqat, and Roohie Naaz Mir. "Evaluation of efficient elastic buffers for the network on
chip router." In 2017 IEEE International Conference on Power, Control, Signals and
Instrumentation Engineering (ICPCSI), pp. 2176-2180. IEEE, 2017.
[7] Aslam, Muhammad Awais, Shashi Kumar, and Rickard Holsmark. "An efficient router
architecture and its FPGA prototyping to support junction based routing in NoC platforms."
In Digital System Design (DSD), 2013 Euromicro Conference on, pp. 297-300. IEEE, 2013.
[8] Kumar, Mayank, Kishore Kumar, Sanjiv Kumar Gupta, and Yogendera Kumar. "FPGA based
design of area-efficient router architecture for Network on Chip (NoC)." In Computing,
Communication, and Automation (ICCCA), 2016 International Conference on, pp. 1600-1605.
IEEE, 2016.
[9] Lu, Ye, John McCanny, and Sakir Sezer. "Generic low-latency NoC router architecture for
FPGA computing systems." In 2011 21st International Conference on Field Programmable
Logic and Applications, pp. 82-89. IEEE, 2011.
[10] Monemi, Alireza, Chia Yee Ooi, and Muhammad Nadzir Marsono. "Low latency network-on-
chip router microarchitecture using request masking technique." International Journal of
Reconfigurable Computing 2015 (2015): 2.

[11] Boutalbi, M., M. Frihi, S. Toumi, C. Tanougast, C. Killian, A. Chaddad, and A. Dandache.
"Reliable router for accurate online error detection in dynamic Network on Chip."
In Microelectronics (ICM), 2013 25th International Conference on, pp. 1-4. IEEE, 2013.
[12] Wang, Wei, Runfeng Li, Fang Fang, Tian Chen, Fuji Ren, Jun Liu, and Xi Wu. "Design and
realization of 3D NOC multicast router base on multicast rotational routing arithmetic."
In Computing, Communication and Networking Technologies (ICCCNT), 2014 International
Conference on, pp. 1-6. IEEE, 2014.
[13] Salah, Yahia, and Rached Tourki. "Design and FPGA implementation of a QoS router for
networks-on-chip." In Next Generation Networks and Services (NGNS), 2011 3rd International
[14] Abdelrasoul, Maher, Mohammed Ragab, and Victor Goulart. "Evaluation of the scalability of
round robin arbiters for NoC routers on FPGA." In 2013 IEEE 7th International Symposium on
Embedded Multicore Socs, pp. 61-66. IEEE, 2013.
[15] Jayan, Geethu, and P. P. Pavitha. "FPGA implementation of an efficient router architecture
based on DMC." In Emerging Technological Trends (ICETT), International Conference on, pp.
1-6. IEEE, 2016.
[16] Liu, Tianqi, Naveen Kumar Dumpala, and Russell Tessier. "Hybrid hard NoCs for efficient
FPGA communication." In Field-Programmable Technology (FPT), 2016 International
[17] Kapre, Nachiket. "Hoplite-Q: Priority-Aware Routing in FPGA Overlay NoCs." In 2018 IEEE
26th Annual International Symposium on Field-Programmable Custom Computing Machines
(FCCM), pp. 17-24. IEEE, 2018.
[18] Choudhari, E. M., and P. K. Dakhole. "Design and verification of five port router for the network
on chip." 2014 International Conference on Communication and Signal Processing. IEEE, 2014.

FPGA IMPLEMENTATION OF PRIORITYARBITER BASED ROUTER DESIGN FOR NOC SYSTEMS

More Related Content

What's hot (18)

Similar to FPGA IMPLEMENTATION OF PRIORITYARBITER BASED ROUTER DESIGN FOR NOC SYSTEMS (20)

More from IAEME Publication (20)

Recently uploaded (20)

FPGA IMPLEMENTATION OF PRIORITYARBITER BASED ROUTER DESIGN FOR NOC SYSTEMS