Research Article
Low-Cost Fault Tolerant Methodology for Real
Time MPSoC Based Embedded System
Mohsin Amin, Muhammad Shakir, Aqib Javed, Muhammad Hassan, and Syed Ali Raza
Department of Electrical Engineering, COMSATS Institute of Information and Technology, Abbottabad, Pakistan
Correspondence should be addressed to Mohsin Amin; mohsinamin@hotmail.com
Received 28 July 2014; Revised 27 October 2014; Accepted 27 October 2014; Published 19 November 2014
Academic Editor: Nadia Nedjah
Copyright © 2014 Mohsin Amin et al. Tis is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
We are proposing a design methodology for a fault tolerant homogeneous MPSoC having additional design objectives that include
low hardware overhead and performance. We have implemented three diferent FT methodologies on MPSoCs and compared
them against the defned constraints. Te comparison of these FT methodologies is carried out by modelling their architectures in
VHDL-RTL, on Spartan 3 FPGA. Te results obtained through simulations helped us to identify the most relevant scheme in terms
of the given design constraints.
1. Introduction
Tere has always been an urge of high performance; subse-
quent to this, plenty of eforts have been made to attain higher
levels in said aspect. In past, technology scaling was mainly
used for performance improvement. Tis approach was quite
helpful in handling various applications with high processing
demands; however bottleneck of this approach has been
reached now [1]. Continuous transistor scaling resulted in
reduced nodal capacitances and low supply voltages making
the devices more prone to the transient faults caused by
external (alpha particles and neutrons) and internal (power
supply noise and cross talk noise) noises [2].
Transient faults are more likely to occur than permanent
faults [3]; that is, product monitoring shows that permanent
faults rate as a result of external events is not more than 10 FIT
(failure in time) whereas the transient fault rate of 1 Mbit of
SRAM is around 1,000 FIT for modern process technologies.
Tus transient faults have major part in disturbing the
reliability of the system [4].
Methods such as rollback are considered efective in
rectifying transient errors. However in hard real time systems
such corrections have a huge infuence on the response time
of the system. A deadline miss in the task can result in the
failure of the system [5]. For example, ATM machines, cell
phones, and thumbprint scanners are some real time systems.
In last decade, to address high performance need there
has been a shif towards multiple resource system. Diferent
interconnect structures have been employed in such systems
including point to point (P2P) and bus based [6]. Both
these architectures were quite useful until we had only
few processors connected through them but as the number
of processing elements (P.E) increased several issues were
spotted and P2P has limited scalability due to increased
complexity [7] while, in buses, scalability is restrained due to
the fact that by raising the number of processing elements
the resulting propagation delay increases [8]. Bus based
architecture can be implemented conveniently if number of
processors are less than fve [9].
To overcome the defciencies of the above mentioned
interconnect structures network on chip (NoC) communi-
cation architecture has been adopted. Tis approach has
the tendency to fulfl high performance requirements and it
provides the designers with several options to optimize their
design according to their requirement [10].
Multiple processors system on chips (MPSoCs) con-
nected through NoC have the ability to perform communi-
cation and computation separately.
MPSoCs featuring NoC can be used to perform various
complicated tasks in parallel such as multimedia streaming,
telecommunication protocols, and GPS [11]. A lot of efort
has been put into the improvement of NoC’s components.
Hindawi Publishing Corporation
International Journal of Reconfigurable Computing
Volume 2014, Article ID 806237, 8 pages
http://dx.doi.org/10.1155/2014/806237