A fault tolerant NoC architecture with runtime adaptive double layer error control and crosstalk avoidance

This paper proposes fault tolerant Network on Chip (NoC) architecture which enables switching of error control coding scheme present in data link layer and network layer as needed, depending upon the rate of error at runtime. The proposed Joint Crosstalk Avoidance-Five Bit Error Correction-Six Bit Error Detection (JCA-FBEC-SBED) error control coding scheme is used in both the layers. This scheme provides crosstalk avoidance and also random and burst error correction up to 5 bits and detection up to 6 bits. The error detection outcomes at all routers in the path are recorded in the error information flit. With the help of error information flit, the error rate is calculated in the destination network interface. The calculated error rate is compared with two threshold values selected based upon the traffic pattern used. If the error rate is less than the lower threshold value, only network layer error control coding scheme is activated. If the error rate is in between the lower and higher threshold values then the error control scheme present in both network layer and datalink layer will be activated, but the error control scheme present in data link layer is activated only in the alternate routers present in the routing path. If the error rate crosses higher threshold value, error control coding scheme present in both the layers will be activated. The proposed JCA-FBEC-SBED error control coding scheme has higher reliability in terms of error detection and correction, when compared to other error control coding schemes with trade-off in delay, area and power consumption. The proposed router architecture has reduced delay and slight increase in area and power consumption of 2.1% and 5.8% respectively, when compared to the runtime adaptive scrubbing router. Our proposed fault tolerant NoC architecture theoretically provide higher data transfer reliability and energy efficiency when compared to other double layer runtime adaptive fault tolerant NoC architectures. The runtime adaptive error control reduces the overall power consumption of the NoC architecture even though JCA-FBEC-SBED decoder consumes more power compared to other schemes.

[1]  Naresh R. Shanbhag,et al.  Coding for systern-on-chip networks: a unified framework , 2004, Proceedings. 41st Design Automation Conference, 2004..

[2]  Cecilia Metra,et al.  Configurable Error Control Scheme for NoC Signal Integrity , 2007, 13th IEEE International On-Line Testing Symposium (IOLTS 2007).

[3]  Kurt Keutzer,et al.  Bus encoding to prevent crosstalk delay , 2001, IEEE/ACM International Conference on Computer Aided Design. ICCAD 2001. IEEE/ACM Digest of Technical Papers (Cat. No.01CH37281).

[4]  Paul Ampadu,et al.  Dual-Layer Adaptive Error Control for Network-on-Chip Links , 2012, IEEE Transactions on Very Large Scale Integration (VLSI) Systems.

[5]  Luca Benini,et al.  Analysis of error recovery schemes for networks on chips , 2005, IEEE Design & Test of Computers.

[6]  Martin Radetzki,et al.  Fault Localizing End-to-End Flow Control Protocol for Networks-on-Chip , 2013, 2013 21st Euromicro International Conference on Parallel, Distributed, and Network-Based Processing.

[7]  G. Seetharaman,et al.  Multi bit random and burst error correction code with crosstalk avoidance for reliable on chip interconnection links , 2013, Microprocess. Microsystems.

[8]  M. Y. Hsiao,et al.  A class of optimal minimum odd-weight-column SEC-DED codes , 1970 .

[9]  Partha Pratim Pande,et al.  Crosstalk-Aware Channel Coding Schemes for Energy Efficient and Reliable NOC Interconnects , 2009, IEEE Transactions on Very Large Scale Integration (VLSI) Systems.

[10]  Wei Hwang,et al.  Self-Calibrated Energy-Efficient and Reliable Channels for On-Chip Interconnection Networks , 2012, J. Electr. Comput. Eng..

[11]  Partha Pratim Pande,et al.  Design of Low Power & Reliable Networks on Chip Through Joint Crosstalk Avoidance and Multiple Error Correction Coding , 2008, J. Electron. Test..

[12]  Avinash Karanth Kodi,et al.  Energy-efficient Runtime Adaptive Scrubbing in fault-tolerant Network-on-Chips (NoCs) architectures , 2013, 2013 IEEE 31st International Conference on Computer Design (ICCD).

[13]  Martin Radetzki,et al.  Fault Tolerant Network on Chip Switching With Graceful Performance Degradation , 2010, IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems.

[14]  Chunjie Duan,et al.  Analysis and avoidance of cross-talk in on-chip buses , 2001, HOT 9 Interconnects. Symposium on High Performance Interconnects.