CPCA: An efficient wireless routing algorithm in WiNoC for cross path congestion awareness

doi:10.1016/j.vlsi.2019.03.008

Integration

Volume 69, November 2019, Pages 75-84

https://doi.org/10.1016/j.vlsi.2019.03.008 Get rights and content

Highlights

•
We established an efficient wireless routing algorithm in WiNoC for cross path congestion awareness (CPCA).
•
In the proposed topology, the wired routers in the same dimension can get the congestion degree of wireless routers.
•
CPCA can compute the optimal path for packets routing, which can prominently avoid the congestion aggravation.

Abstract

Wireless network-on-chip (WiNoC) is a new paradigm to mitigate the long-distance transmission latency for conventional wired network-on-chip. The wireless routers in WiNoC have to handle a large number of packets which could cause data congestion, thus reducing the network performance. In this paper, we propose a novel wireless routing algorithm, called CPCA, which exploits the cross path congestion information as hints to route the packets. Under CPCA, the whole network is partitioned into sub-networks. In each subnet, the congestion information of the wireless router is propagated along the cross path. As a result, the routers in the same dimension can get the congestion degree of wireless router within the subnet. Based on the congestion information, CPCA can compute the suitable path for packets routing, which can prominently avoid the congestion aggravation in the wireless router. Experimental results show that our proposed method can effectively improve performance in terms of packets transmission latency and network throughput.

Introduction

Network-on-Chip (NoC) is an emerging interconnection paradigm for the multi-core processors. However, as more cores are integrated into chips, conventional NoC suffer from high latency and high power consumption [[1], [2], [3]]. According to ITRS [4], new interconnection technology should be developed to further improve performance. Recently, several novel technologies, such as 3D NoC, optical NoC (ONoC) and RF interconnection (RF-I) have been proposed [[5], [6], [7], [8]]. For 3D NoC [5], chips are vertically stacked together which greatly reduces the routing distance between cores. ONoC [7] outperforms because of its excellent characteristics such as high bandwidth and low electromagnetic interference. For RF-I [8], data in the transmission line can be transferred to the destination approaching the speed of light. However, these three technologies have some technical or practical challenges which should be addressed [9], such as complexity and thermal regulation.

Through replacing baseline wired routers with wireless routers (WRs), WiNoC can achieve long-distance data transmission through one-hop wireless link [10,11]. Compared with wired NoC, WiNoC supports more flexible routing strategies. Moreover, WiNoC is compatible with existing CMOS process which makes its easy on-chip integration [12]. However, WiNoC also has some challenges, such as network congestion [13]. As an important part of WiNoC, wireless router not only take the responsibility to forward the short distance packets, but also transmit long distance packets. It is possible that a large number of packets can be flooded into the wireless router which will lead to network congestion and thermal hotspot [14]. Network congestion directly reduces the network throughput, thus degrading system performance. As a result, it is imperative to find a proper scheme to alleviate network congestion, so as to improve the system performance.

In this paper, we propose an efficient WiNoC routing algorithm called CPCA. Through the congestion information of the wireless router within the subnet, the CPCA algorithm can make near-optimal selection of wired links and wireless links during packet routing, so as to mitigate the congestion of wireless router. The experimental results show the performance benefits clearly outweigh the overheads. The main contributions of this paper include the following:

1)
Through partitioning the network into sub-networks, we present a novel mechanism to propagate congestion information. In one subnet, the 1-bit signal line is formed as cross path. Using 1-bit signal line, this mechanism can transmit the congestion information of the wireless router along the cross paths within the subnet, so that the routers along the paths can perceive the congestion information of wireless router.
2)
We propose an effective wireless routing algorithm which makes near-optimal selection of wired links and wireless links during packet routing.
3)
We conduct extensive experiments to evaluate the performance of the proposed algorithm against other emerging schemes, under different traffic patterns. We also evaluate the area overhead and power consumption of different schemes. Our experimental results show that the proposed scheme can offer significantly better performance in malterms of transmission latency, throughput and packet energy compared with other counterparts.

The remainder of this paper is organized as follows. Section 2 presents the background and related work. In Section 3, the proposed routing scheme is presented. The experiments as well as evaluation are presented in Section 4. Section 5 concludes the paper.

Section snippets

Background and related work

In this section, we present several popular wireless communication technology in WiNoC. We also show some extensive research on alleviating the congestion in WiNoC.

Proposed cross path congestion awareness routing

In this section, we first present the WiNoC architecture including the topology and router architecture. Subsequently, we describe the proposed routing algorithm and demonstrated using example in detail.

Experiments

In this section, we analyze the characteristics and performance of different schemes under different traffic patterns. We first present the simulation environment in Section 4.1. Subsequently, we simulate and analyze the experimental results in detail.

Conclusion

As a novel on-chip interconnect architecture, WiNoC can break the performance limit of traditional wire line counterpart. Through the using of wireless links, the packets can achieve one-hop transmission between long distance nodes. However, excessively using wireless short-cut may occurs congestion which reduces communication efficiency of the network. In this paper, we propose an efficient wireless routing algorithm for cross path congestion awareness. Based on the proposed topology, the

Acknowledgments

This research is supported in part by the National Natural Science Foundation of China (NSFC) research Projects 61474036, 61274036 and 61371025, the Natural Science Foundation of Anhui Province China under the Grant 1508085MF117.

References (38)

M.S. Rahaman et al.
Energy efficiency of error control coding in intra-chip RF/wireless interconnect systems
Microelectron. J.
(2010)
A. Mineo et al.
Exploiting antenna directivity in wireless NoC architectures
Microprocess. Microsyst.
(2016)
P. Dai et al.
A study of a wire–wireless hybrid NoC architecture with an energy-proportional multicast scheme for energy efficiency
Comput. Electr. Eng.
(2015)
P.J. Hsu et al.
The design and implementation of a configurable MIMO detection system on the NOC-based multicore platform
Microelectron. J.
(2016)
L. Wang et al.
A runtime fault-tolerant routing algorithm based on region flooding in NoCs
Microprocess. Microsyst.
(2016)
P.P. Pande et al.
The (low) power of less wiring: enabling energy efficiency in many-core platforms through wireless noc
Linda Wilson
International Technology Roadmap for Semiconductors (ITRS)
(2013)
K.C. Chen et al.
Topology-aware adaptive routing for nonstationary irregular mesh in throttled 3D NoC systems
IEEE Trans. Parallel Distrib. Syst.
(2013)
L. Zeng et al.
An efficient highly adaptive and deadlock-free routing algorithm for 3D network-on-chip
IEICE Trans. Fundam. Electron. Commun. Comput. Sci.
(2016)
S. Koohi et al.
An optical wavelength switching architecture for a high-performance low-power photonic noc

M. ValadBeigi et al.

An energy-efficient reconfigurable NoC architecture with RF-interconnects

A. Ganguly et al.

Performance evaluation of wireless networks on chip architectures

S. Deb et al.

Design of an energy-efficient CMOS-compatible NoC architecture with millimeter-wave wireless interconnects

IEEE Trans. Comput.

(2013)

C. Wang et al.

A wireless network-on-chip design for multicore platforms

A. Karkar et al.

A survey of emerging interconnects for on-chip efficient multicast and broadcast in many-cores

IEEE Circuits Syst. Mag.

(2016)

L. Wang et al.

Flow control mechanism for wireless network on-chip

J. Murray et al.

Thermal hotspot reduction in mm-wave wireless noc architectures

K. Duraisamy et al.

Multicast-aware high-performance wireless network-on-chip architectures

IEEE Trans. Very Large Scale Integr. Syst.

(2017)

S. Deb et al.

Wireless NoC as interconnection backbone for multicore chips: promises and challenges

IEEE J. Emerg. Select. Topic. Circ. Syst.

(2012)

Cited by (7)

Design of fully adaptive routing and hybrid VC allocation in wireless NOC
2022, Microelectronics Journal
Citation Excerpt :
The deadlock problem in WiNoC is further complicated by the addition of long-range wireless links. The Dataline scheme, which is used in the Torus network to solve long-range links, has been widely used in WiNoC [6,8–12]. The Dataline solution is simple and effective, but requires at least two VCs (Virtual Channels) to avoid deadlocks.
Wireless network-on-chip (WiNoC) establishes a long-range single-hop wireless links on the base of the traditional planar metal interconnects network, which improve network performance in several aspects. The establishment of wireless link improves the diversity of path for packet transmitting, while simultaneously aggravates the deadlock problem, leading it to more complicated condemn. The traditional routing mechanism of wireless network on chip can avoid deadlock by sacrificing the adaptability of transmitting, which leads to unbalanced and insufficient use of limited on chip resources. In this paper, a fully adaptive routing mechanism is proposed to balance buffer utilization. At the same time, we proposed a hybrid virtual channel allocation strategy in order to make full use of buffer resources. For a large number of single-flit wired packets and wireless packets, more aggressive non-atomic allocation is adopted, and the remaining packets are arranged with atomic allocation to avoid potential deadlock. Experimental results showed that compared with the traditional deadlock solutions, the proposed scheme introduces less area and power consumption, and balances the utilization of buffer resources. Under the non-uniform Traffic Patterns, the proposed scheme has better performance in terms of delay and throughput, achieves 62.0% improvement compared to the traditional counterpart.
MMNNN: A tree-based Multicast Mechanism for NoC-based deep Neural Network accelerators
2021, Microprocessors and Microsystems
Citation Excerpt :
A NoC is a type of packet-switched network that is composed of routers and links. Significant research has been made into both wired [3–5] and wireless NoCs [6–9]. In a typical NoC, each router is connected to a processor element (PE) or group of PEs, with routers connected to each other by links [10].
Network-on-Chip (NoC) devices have been widely used in multiprocessor systems. In recent years, NoC-based Deep Neural Network (DNN) accelerators have been proposed to connect neural computing devices using NoCs. Such designs dramatically reduce off-chip memory accesses of these platforms. However, the large number of one-to-many packet transfers significantly degrade performance with traditional unicast channels. We propose a multicast mechanism for a NoC-based DNN accelerator called Multicast Mechanism for NoC-based Neural Network accelerator (MMNNN). To do so, we propose a tree-based multicast routing algorithm with excellent scalability and the ability to minimize the number of packets in the network. We also propose a router architecture for single-flit packets. Our proposed router transfers flits to multiple destinations in a single process and has no head-of-line blocking issue, offering higher throughput and lower latency than traditional wormhole router architectures. Simulation results show that our proposed multicast mechanism offers excellent performance in classification latency, average packet latency, and energy consumption.
A survey and taxonomy of congestion control mechanisms in wireless network on chip
2020, Journal of Systems Architecture
Citation Excerpt :
It can be seen that LTCA is an efficient congestion control scheme that does not increase hardware resources, while it can be reduceed the congestion of WRs and can be improvd average latency. Ouyang et al. [30] proposed CPCA; is a new congestion-aware routing algorithm that uses the cross path congestion information for packets routing. To detect congestion in a subnet, WR congestion information is propagated among BRsthat are in the cross path with WR.
Wireless network on chip (WiNoC) has been proposed as a promising solution for on-chip interconnection network due to high scalability, high bandwidth, and low latency. However, the variations of traffic pattern distribution and data flow lead to congestion in wireless interfaces-equipped routers (WRs). Congestion is one of the main challenges in emerging WiNoCs that can reduce network performance. In recent years, various researches have been proposed to decrease congestion in WiNoCs. However, this paper presents a comprehensive survey of the significant congestion control mechanisms in WiNoCs. The available schemes are classified into six categories, includeing hardware resources-based congestion control, congestion-aware routing algorithms, medium access control protocol,congestion-aware architectures, rate-based congestion control, and application-mapping with task-migration techniques. The goal of this survey is to highlight the characteristics and inherent constraints of congestion control mechanisms in a novel approach that can help researchers for designing efficient congestion control scheme.
On the Impact of Subnet Clustering in Radio Hub for 100-core Wireless Network-on-Chip Architecture
2024, International Journal of Computing and Digital Systems
Performance Evaluation of Multi-Channel for 10×10 Mesh Wireless Network-on-Chip Architecture
2022, 2022 IEEE International Conference on Computing, ICOCO 2022
Energy-efficient networks-on-chip architectures: Design and run-time optimization
2021, Network-on-Chip Security and Privacy

View all citing articles on Scopus

View full text