DEVELOPING SCHEDULER TEST CASES TO VERIFY SCHEDULER IMPLEMENTATIONS IN TIMETRIGGERED EMBEDDED SYSTEMS

International Journal of Embedded systems and Applications (IJESA) Vol.6, No.1/2, June 2016
DOI: 10.5121/ijesa.2016.6201 1
DEVELOPING SCHEDULER TEST CASES TO VERIFY
SCHEDULER IMPLEMENTATIONS IN TIME-
TRIGGERED EMBEDDED SYSTEMS
Mouaaz Nahas1
and Ricardo Bautista-Quintero2
1
Department of Electrical Engineering, College of Engineering and Islamic Architecture,
Umm Al-Qura University, Makkah, KSA
mmnahas@uqu.edu.sa
2
Department of Mechanical Engineering, InstitutoTecnólogico De Culiacán, Sinaloa,
México
r.bautista@unb.ca
ABSTRACT
Despite that there is a “one-to-many” mapping between scheduling algorithms and scheduler
implementations, only a few studies have discussed the challenges and consequences of translating between
these two system models. There has been an argument that a wide gap exists between scheduling theory and
scheduling implementation in practical systems, where such a gap must be bridged to obtain an effective
validation of embedded systems. In this paper, we introduce a technique called “Scheduler Test Case”
(STC) aimed at bridging the gap between scheduling algorithms and scheduler implementations in single-
processor embedded systems implemented using Time-Triggered Co-operative (TTC) architectures. We will
demonstrate how the STC technique can provide a simple and systematic way for documenting, verifying
(testing) and comparing various TTC scheduler implementations on particular hardware. However, STC is
a generic technique that provides a black-box tool for assessing and predicting the behaviour of
representative implementation sets of any real-time scheduling algorithm.
KEYWORDS
Scheduler algorithm, scheduler implementation, cyclic executive, time-triggered co-operative scheduler,
resource-constrained embedded system, scheduler test cases, predictability, jitter, task overrun.
1. INTRODUCTION
There are numerous ways in which we can describe (and distinguish) the different architectures
employed in embedded computer systems. The architecture which forms the focus of this paper is
usually described as “time triggered” (TT); as opposed to “even-triggered” (ET) [1]. For an
embedded system with a time-triggered architecture, we are supposed to know in advance how
the system will behave exactly at every instance during its running time. Knowing the complete
behaviour of the system and hence determining whether the system is capable of meeting all its
timing constraints can be referred to as “predictability” [2], [3].
Such an ideal TT behaviour is inevitably hard to achieve in real world. However, approximations
of this model have been developed and widely used in practice. The best approximation of an
ideal TT architecture contains a set of periodic tasks running in a co-operative (or simply
“nonpre-emptive”) manner. Such a design is referred to as “time-triggered co-operative” (TTC)
architecture [4]–[8]. Early versions of this architecture used to be called a “cyclic executive” [9]–
[11]. Unlike time-triggered pre-emptive algorithms (e.g. “rate monotonic”), systems with TTC
architectures have high predictability in their timing behaviour, i.e. they demonstrate very low
levels of task jitter [10] and have the ability to maintain such low-jitter characteristics even when

2
modern techniques are integrated to improve system performance or minimise power
consumption (e.g. “dynamic voltage scaling” (DVS) technique [12]).
The principal aim of this paper is to ensure that accurate timing predictions made at the design
phase are maintained during the process of implementing (and also maintaining) a practical
system. It has been argued that there is a “one-to-many” mapping between scheduling algorithms
and scheduler implementations ([9], [13]–[15]). However, the process of transforming between
these two system representations has not received widespread attention. In an effort to facilitate a
systematic approach for validating an implementation of TT system, we introduce an empirical
approach which we call a “scheduler test case” (STC). Such a technique is not intended to test all
features of the system (by any means): instead, it is intended to be used with “dummy tasks”
solely as a practical means for assessing the scheduler implementation employed in a given
system. In this way, we aim to allow those implementing the system to gain a better
understanding (during system construction, testing or maintenance) of the way in which a given
TTC implementation is expected to behave under a range of normal and abnormal operating
conditions. Our objective is that the developed STC method can provide a “black box” evaluation
of a given scheduler implementation without the necessity to access (or even comprehend) the
underlying source code by the system developer (or user).
Predictability is an important design parameter in real-time resource-constrained embedded
systems, particularly in those employing TTC architectures. Thus, it is used here as the essential
criteria for assessing the behaviour of a scheduler. To be able to express predictability using
qualitative and quantitative measures, the following three criteria are considered: (1) task
execution sequence; (2) timing jitter; and (3) the capability of the scheduler to handle unexpected
errors. In addition, computational, memory and power overheads resulted from the
implementation of each scheduler are used to allow a practical comparison between the various
studied schedulers.
The remainder of the paper is organised as follows. In Section 2, we provide a detailed literature
review of the work conducted in the same field of this study. In Section 3, we introduce what we
call “Scheduler Test Cases” (STCs). In Section 4, we demonstrate how the STC technique is
employed in TTC algorithm. In Section 5, the methodology used to attain the empirical results of
this study is described. In Section 6, we report the results obtained when the STC is applied to a
set of representative TTC scheduler implementations collected from the literature. Finally, we
present our conclusions in Section 7.
2. LITERATURE REVIEW
In this section, a detailed literature review of the work carried out in this area is provided.
2.1. Automated code generation
One way to bridge the gap between scheduler algorithms and scheduler implementation is by
means of automated code generation. Such techniques help to reduce the time and effort
consumed in the implementation process of safety-critical systems, while removing errors that are
likely to arise during this stage of development [16], [17]. Various industries – such as aerospace
and automotive – have extensively used automatic code generation tools for control and signal
processing systems [18]–[20]. TTC systems considered in this study are typically used in such
application arenas. It is believed that hundreds of thousands of modern cars rely on codes
generated using automatic code generation [18], [21].

3
Previous work on the “automatic” generation of systems using TTC architecture are presented in
[22], [23]. This kind of work helps the developer to utilise a collection of “design patterns” for
creating a complete code (with the underlying scheduler) of a TTC system. In addition, even with
TTC architectures, the user still needs to “hand tune” some task parameters (like the offset) and
scheduler parameters (like the tick interval). Incorrect tuning of these parameters may result – at
worst – in unschedulability of tasks. However, even if all tasks are ensured to be scheduled,
incorrect decisions may still cause degradation to the system performance (e.g. introducing high
levels of task jitter) or an increase in the implementation costs (e.g. increasing CPU overhead or
system power consumption). To conclude, automated code generation techniques – despite their
important role in validating embedded software – do not take into account the possible runtime
behaviour that can be resulted from a particular implementation of software code. For example,
no feedback process is incorporated in such techniques which allows the designer to examine the
consequence of using a particular code on the runtime behaviour of their system and hence
modify the source code accordingly.
2.2. Scheduler implementation
It is noticed that a great deal of work has been carried out in developing, assessing and modifying
(enhancing) scheduling algorithms. Despite that, only a limited amount of work has been found
on the process of scheduler implementation and its vulnerability to the decisions made at early
stages of the development lifecycle (namely, during scheduler design). The relationship between
any scheduling algorithm and its possible implementation options is usually viewed as “one-
tomany” [4], [9], [14], [24]–[26], which simply means that the output behaviour of the system
would entirely depend on the implementation decisions. To elaborate more on this point, if the
source code of a scheduler is implemented using an appropriate collection of “software design
patterns” which is a common practice [4], the relationship between a design pattern and its
implementation is typically described as “one pattern, many implementations” [27]. This would
inevitably cause the scheduling algorithm to have a wide range of scheduler implementations;
each corresponds to a different pattern. Figure 1 shows the implementation of a simple TTC
algorithm based on a set of different software design patterns, where each pattern represents a
different scheduler implementation (e.g. TTC-1, TTC-2, etc.) and uses various Pattern
Implementation Examples (PIEs) corresponding to different processor platforms (e.g. 8-bit 8051,
16-bit c167 or 32-bit ARM microprocessor). This figure helps to explain the concept behind the
“one-to-many” relationship between a particular scheduling algorithm and its practical
implementations in real-time, resource-constrained embedded systems (see [27] for more details).
Figure 1. The “one-to-many” mapping between the TTC scheduler and its practical implementations using
software design patterns (This figure is reproduced from [27]).

4
The optimum schedule for a set of real-time tasks is determined. In contrast, the latter term
describes the process in which the scheduler is physically implemented – either in software or
hardware – to execute the designed task-schedule at the system operating time [28]. An early
work considering the implementation of TTC architectures using the Ada programming language
was conducted in [9]. Later, the work in [29] considered the issues of implementing forms of
cyclic executive (i.e. a simple TTC algorithm) in Assembly language. The work outlined in [12]
developed techniques to maintain low task jitter when dynamic voltage scaling is applied in TTC
algorithm to reduce power consumption of the system. In our previous work [30], [31], we
developed various low-jitter TTC schedulers aimed at highly-predictable embedded applications.
In [6], [32], [33], we developed and evaluated low-jitter TTC schedulers for multi-processor
distributed real-time embedded systems built using Controller Area Network (CAN) protocol.
Various ways in which TTC algorithm is implemented in multi-processor embedded systems were
described in detail in [34]. The potential impact of scheduler implementation on task jitter for
TTC algorithm in both single- and multi-processor designs was studied in detail in [15]. In [35],
[36], TTC schedulers with “task guardian” mechanisms were introduced to address the task
overrunning problem.
In trying to link scheduling algorithm and scheduler implementation, Katcher et al. [13] argued
that a wide gap exists between the theory of scheduling and its practical implementation in
operating system cores running on specific hardware platforms. Moreover, they asserted that
bridging such a gap is imperative if a meaningful validation of real-time systems is sought to be
achieved. The same argument was made by other researchers (e.g. [37], [38]). For example, [13]
stated that the implementation of a particular scheduler may introduce costs that must not be
overlooked if the timing behaviour of a real-time system is to be validated.
2.3. Software verification and testing
Since our focus is to link scheduling algorithms and scheduler implementations, we should think
of a way to ensure a complete matchup (i.e. compliance) between the design and implementation
processes of the scheduler (with a particular focus on software implementation). Software
engineers refer to this process as “software verification”. We have found evidence that confusion
exists in using the terms “validation” and “verification” by many people concerned with the
development and evaluation of software applications. For example, some people think that
“validation” is a synonym to “verification” and vice versa [39], and hence use the two terms
interchangeably. Based on a wide range of definitions collected from various dictionaries, a
system is considered to be validated (or valid) if its final software product fulfils the requirements
of the user (or the customer who pays for it). Therefore, any process involved in testing such a
fulfilment is called “validation process”. In contrast, verification requires checking each
component in the system to ensure that its output matches the specifications of its input and hence
maintains the integrity of the user’s initial requirements [2]. In [40], it is noted that “validation
usually takes place at the end of the development cycle, and looks at the complete system as
opposed to verification, which focuses on smaller sub-systems”. Figure 2 shows one possible way
in which we attempt to integrate validation and verification processes in the general model of
software development life cycle as illustrated by [39].
Figure 2. Inserting validation and verification processes in the software development life cycle

5
For system verification, there are two main approaches: “static” and “dynamic”. Static approach
involves software inspections and formal methods, while dynamic one mainly involves software
testing [39]. Among these techniques, software testing scores the best in verifying the
implementation of a given real-time scheduler, since it allows testing dynamic features which
only manifest when the system is operational (this is an important requirement in real-time system
where correct timing of results is crucial). A detailed comparison between the various verification
techniques is provided in [2]. For any testing process, an appropriate set of test cases has to be
designed. In fact, only an effective subset of possible test cases is used. A general model for
software testing process is illustrated in Figure 3. The figure reveals the basic elements of any
testing process which are: (1) test cases; (2) test data (i.e. test input); and (3) test output. The test
output has to be compared with the output estimated at the design process of the test case by
developers who possess a complete knowledge about the system and its expected operation at
real-time.
Figure 3: Model of testing process (reproduced from [39])
2.4. Test cases and test-case generation
Many researchers have argued that testing may consume almost half of the total development cost
[39], [41]–[46]. Lots of studies have therefore demonstrated that test process automation holds the
promise to reduce time, effort and costs (e.g. [47]–[49]). As mentioned in the introduction, the
focus in our paper is on a form of test cases.
There has been a great deal of previous work on both test cases and test-case generation. For
example, Beck’s work on “extreme programming” has at heart a view that test cases for the
system should be produced early in the product life cycle [50]. In [51], it is argued that the most
challenging phase in the test process is the selection and execution of test cases. Many studies
have, therefore, proposed techniques for automatic test-case generation (see for instance, [43],
[47], [49], [51]). Test cases may be created directly from the source code, or other sources such as
control flow graphs, design representations and specifications [52]. Many studies have considered
the process of generating test cases for real-time computing systems (e.g. [47], [51], [53]–[57]).
In this section, only a small collection of literature has been reviewed to show the interest of
researchers in this area. In effect, previous work on testing considered the examination of
software functionality or software quality attributes (features). The authors of this paper are not
aware of any extensive work which considered the development of test cases solely to study the
impact of using a particular source code (i.e. scheduler software) on the output behaviour of the
system executing the code, especially when algorithms such as TTC schedulers are implemented.

6
3. SCHEDULER TEST CASE (STC) TECHNIQUE
As noted before, testing here is not aiming to check the correct functionality of the application
software or evaluate its quality attributes. Instead, it is mainly used to evaluate the runtime
behaviour of the system as a result of employing a particular software implementation of TTC
scheduler using commonly-used processor platforms. As in all testing processes, a suitable
collection of test cases are required. Such test cases typically determine the system inputs,
predicted outputs, and execution conditions, and aimed to verify the system conformity with
specific requirements. Again, note that only a selective subset of possible test cases can be used
since inclusive testing is not possible. The feature of the system to be tested must be selected
along with the inputs that will execute that feature, and the expected outputs of the test cases must
be known in advance. All abovementioned test case components have been integrated in the
process of developing the STCs presented here. This is further described as follows.
The STC is a simple technique which employs a collection of test cases to examine the output
behaviour of a wide range of TTC scheduler implementations. The STCs developed here are
created manually depending on previous experience and full comprehension of the characteristics
and constraints of the TTC algorithm [4]. The STC technique employs various scheduling
examples using non-real (dummy) tasks where those examples show a different set of behaviour
patterns as the scheduler implementation varies. Such task sets represent the test inputs (i.e. test
data). Each STC contains a different collection of tasks with different properties. The test item
here is the TTC scheduler to which the tasks in each STCs are entered. The scheduler will then
execute on the target microcontroller hardware for a sufficient period of time during which the
system response is monitored. Then, the output behaviour will be recorded and compared to the
predicted behaviour (which was fully documented at the test case design stage). Figure 4 shows
the detailed process of STC testing developed in this study.
Figure 4. Testing process model for the STC technique (this is extracted from the process
Illustrated in Figure 3)
The STC technique tests the system performance under normal and abnormal operating
conditions. By normal conditions, we mean that the scheduler operates in non-existence of errors.
In contrast, abnormal conditions refer to the situation where errors occur. The main criterion to
define an error mode is that it should represent a recognised problem facing the developers of the
system under verification. For example, TTC systems (which form the key focus of this study)
suffers from “task overrunning” problem which can deteriorate the overall system performance if
not addressed immediately (sometimes, it can spoil the system’s functionality) [35], [36].
Accordingly, task overrunning problem has been selected to represent the error mode in TTC
schedulers.

7
As noted above, the key criteria used to assess the behaviour of each TTC scheduler are: task
sequencing, jitter levels and the ability to address task overrunning problem. You can look at
these criteria as the main tested features of the TTC scheduling algorithm. Such criteria are used
to reflect the level of predictability in the TTC algorithm. The task sequencing criterion checks if
the scheduler runs tasks in the required order as determined in the schedule design process. Jitter
in the task timing is used to assess the timing performance of the system. In our study, the jitter is
measured at the release time of all tasks running in the TTC scheduler. Release jitter describes the
deviation of the start time of a task from its release time. It can easily be shown that – in many
real-time embedded system designs – reducing jitter on all (or some) tasks has the potential to
increase the overall system’s predictability.
4. THE SCHEDULER TEST CASES (STCS) FOR TTC ALGORITHM
In this section we present the various “scheduler test cases” (STCs) developed in this study to
analyse and test the behaviour of different implementation classes of the TTC scheduler.
4.1 STC A (Task-induced jitter)
STC A investigates the impact of variations in the task execution time on the jitter levels. Figure 5
illustrates the way in which a particular TTC implementation might be expected to execute the set
of tasks which we consider in STC A. In this figure (and others in this paper) the vertical arrows
represent the points at which periodic timer interrupts (or “ticks”) occur. The interval between
successive timer interrupts is known as “tick interval”. In Figure 5, all tasks execute with a “tick
offset” value of 0. This simply means that each task runs for the first time in tick interval number
zero.
Task characteristics for this STC are detailed in Table 1. Considering STC A in more detail, the
figure illustrates how Task B (and Task C) may suffer from release jitter (since the execution
duration of Task A varies from one tick to another).
Figure 5. Graphical representation of Example schedule A1
Table 1: Task set for STC A (Major cycle = 1 Tick).

8
Examples of possible schedules obtained with this task set are given in Table 2 and Table 3.
Table 2: Example schedule A1.
Comment:
In a simple scheduler implementation, it is expected to see high levels of jitter in the start times of
tasks executed later in the tick interval, if the execution time of the earlier tasks is variable. This is
illustrated in Figure 5. Note that such a scheduler implementation is not suitable for use with
jitter-sensitive tasks
Table 3: Example schedule A2.
Comment:
In a scheduler with low-jitter characteristics, the scheduler can compensate for variations in the
execution time of tasks. Thus, tasks with lower priorities will not suffer from high jitter at their
release times. This is illustrated in Figure 6.
Figure 6. Graphical representation of Example schedule A2
4.2 STC B (Schedule-induced jitter)
Unlike STC A, STC B investigates the impact of variations in the schedule on the jitter levels of
tasks. A summary of the task characteristics for this STC is presented in Table 4.
Table 4: Task set for STC B (Major cycle = 2 Ticks).

9
Examples of possible schedules obtained with this task set are given in Table 5 and Table 6.
Table 5: Example schedule B1 (Basic scheduler).
Comment:
Here, the release jitter in Task B is expected to be high as it executes immediately after Task A,
where Task A has a variable duration time. This is shown in Figure 7.
Figure 7. Graphical representation of Example schedule B1
Table 6: Example schedule B2 (TTC scheduler with gap insertion).
Comment:
This scheduler implementation fulfils the jitter requirements for Task B and Task C by
compensating for the variation in the schedule itself (see Figure 8).
Figure 8. Graphical representation of Example schedule B2
4.3 STC C (Long tasks)
TTC scheduler is typically based on the use of timer interrupt that overflows periodically to
generate tick intervals. In STC A and STC B, we assume that all tasks which begin execution in a
given tick interval are intended to complete their execution before the next tick occurs. Such a
restriction is not a fundamental requirement in TTC systems3, but may cause a limitation in some
TTC scheduler implementations. Therefore, we test the scheduler ability to handle “long tasks” in

10
STC C. A summary of the task characteristics for this test is presented in Table 7. Examples of
possible schedules obtained with this task set are given in Table 8 to Table 12.
Table 7: Task set for STC C (Major cycle = 4 Ticks).
Comment
In this sequence, Task B runs for 2.4 ticks. While executing Task B, Task A (which has a low-
jitter requirement) becomes ready to execute. With STC C, we can check how the scheduler will
deal with tasks which are deliberately set to run beyond the tick interval. This test also checks
how the scheduler will manage the task priorities; for example, Task A has a higher priority than
Task C so once Task B finishes executing, Task A should be executed prior to Task C.
Table 8: Example schedule C1 (Basic scheduler)
Comment
In a basic TTC implementation, all delayed tasks will execute in their predefined order upon the
completion of the long task; that is Task B in our case (Figure 9).
Figure 9. Graphical representation of Example schedule C1
Table 9: Example schedule C2.
Comment
In many TTC implementations, the schedule checks the status of the task next to the long one in
the task sequence. If it is due to run, the schedule will execute it regardless of whether higher
priority tasks are waiting to execute or not. In the example shown in Figure 10, Task C will be
executed immediately after Task B and before Task A (which has the highest priority in the whole
task list).

11
Comment
In each major cycle, the first execution of both Task A and Task B is omitted (see Figure 11).
Comment
Here, Task A will be first executed in the last tick of the whole cycle, as depicted in Figure 12.
Comment
The scheduler enforces any long task to shut down as soon as the next tick arrives (Figure 13).

12
4.4 STC D (Task overruns)
All previous STCs assume that the system operates normally without errors. STC D is specifically
designed to investigate the impact of unplanned task overruns. Task characteristics for this test are
presented in Table 13. Examples of possible schedules obtained with this task set are given in
Table 14.
Table 13: Task set for STC D (Major cycle = 20 Ticks).
Comment
In this sequence, Task A is designed to run for the duration of 0.2 Tick. Due to an error, Task A
overruns by 10 Ticks. This is illustrated in Figure 14.
Figure 14. Summary of STC D (Task A overruns by 10 ticks)
Table 14: Example schedule D1a, D1b, D2a, D2b, D3a and D3b.

13
4.5 CPU, memory and power overheads
This paper is concerned with the type of embedded systems that can be built on “commercial
offthe-shelf” (COTS) microcontrollers. Such small and cheaply-available hardware usually have
limited resources (i.e. CPU, memory and power). We will therefore report CPU, memory and
power requirements (overheads) for all schedulers considered in this paper. Note that in many
embedded system applications, average power consumption is a key concern, as this is related to
the system battery lifetime.
5. METHODOLOGY
In this section, we provide an overview of the methodology used to obtain the empirical results
presented in this paper.
5.1 Representative examples of TTC implementations
In fact, it is impossible to cover all possible implementation options for a simple TTC scheduler
in a single study. Hence, we apply the proposed STC technique on a range of TTC schedulers
which form representative examples of the broad range of TTC implementations developed
previously. The selected implementations are all based on the use of periodic timer interrupts4
and have been gathered from a wide range of different projects carried out in the Embedded
Systems Laboratory (ESL) at the University of Leicester, UK. Such TTC implementations are:
TTC-ISR [4], TTC-Dispatch [4], [32], TTC-DVS [12], TTC-TG [36], TTC-ISR [30] and
TTCAdaptive [60] schedulers. For further details about these schedulers, please refer to the
references listed in this section.
5.2 Hardware platform
As noted before, the work conducted here is based on the use of low-cost embedded systems
based on small microcontroller hardware such as: 8051, Infineon C16x, Philips LPC2xxx, or PH
Processor. For programming the hardware (i.e. writing the software program including the
scheduler code), the C programming language has been found to be competitive among many
other surveyed languages (see [61] for more details).
The empirical studies presented here were carried out using Ashling LPC2000 evaluation board
with built-in Philips LPC2106 processor [62]. The LPC2106 is 32-bit microcontroller with an
ARM7 core that uses an on-chip “phase-locked loop” (PLL) to run the processor at frequencies
ranging from 12 MHz to 60 MHz [63]. The oscillator and CPU frequencies used throughout this
study (except for TTC-DVS) are 12 MHz and 60 MHz, respectively. The compiler used is the
GCC ARM 4.1.1 where the used simulator is the Keil ARM development kit (v3.12).
5.3 Scheduler behaviour test
In each scheduler implementation, the scheduler behaviour (e.g. task sequencing) was measured
directly from the simulator by using breakpoints in each task to observe the order (and the time) at
which the tasks execute. The measurements were made over a number of successive major cycles.
The results obtained when executing each STC were then reported and compared with the
example schedules discussed in Section 4.

14
5.4 Jitter tests
To obtain a meaningful set of task jitter results, Task A, Task B and Task C had variable
durations in STC A and STC B. This is in order to facilitate a detailed study of the impact of
varying task duration on the release jitter of other tasks scheduled in the same ticks. In STC C, we
explored the impact of long tasks on the tick and tasks jitter. It should be noted that the jitter
levels were only considered when the scheduler operates in normal conditions (i.e. jitter levels
were not assessed with STC D which is originally designed to test the system under abnormal
conditions).
To measure the jitter on the tick and tasks using practical means, we set a pin high at the start of
the tick or task (for a short period of time) and then measured the periods between each two
consecutive rising edges. We recorded 5000 samples in each experiment. The periods were
measured using a National Instruments data acquisition card ‘NI PCI-6035E’ [64] along with
LabVIEW 7.1 software [65]. To assess the jitter levels, two values can be considered: “average
jitter” and “difference jitter”. The difference jitter is the difference between the maximum and the
minimum period in the measurements. This kind of jitter is also described as “absolute jitter”
[66]. The average jitter is taken as the standard deviation from the average period. Here, we
reported the difference jitter only. Of course, there can be many other ways to measure jitter in
practice, but such methods have been found to be appropriate for our study.
5.5 CPU, memory and power test
The CPU overhead is one of the cost parameters that have been used here to differentiate between
the various scheduler forms. To get practical CPU overhead measurements for each scheduler, we
ran the STC A for the interval of 25 seconds and measured the total time required to run the
scheduler by the performance analyser tool in the Keil simulator. The CPU overhead results were
then presented in percentages of the total computational resources.
For memory requirements, we recorded the code (ROM) and data (RAM) memory values
required to implement STC A for each scheduler considered in this study. We performed this task
by using the “.map” file created when the STC is compiled. For data memory overhead (i.e. RAM
requirements), we also measured the STACK usage by filling the data memory with a dummy
code and recording the number of changed (overwritten) bytes after the system is executed for a
sufficient period of time.
In order to obtain representative results of power consumption, we measured the input current and
voltage to the LPC2106 CPU core while executing STC A and STC B. The sampling frequency of
10 kHz was used over 5000 major cycles (the need for this specific frequency is explained more
in [12]). The values of currents and voltages were then multiplied and averaged out.
6. RESULTS
In this section, we provide the output results from the application of the STC technique on all
selected TTC implementations. As we have attempted to make clear, even the small (and by no
means exhaustive) selection of TTC scheduler implementations demonstrate a wide range of
different patterns of behaviour, and a “one size fits all” implementation does not exist. We
suggest that the results obtained by applying the scheduler test cases (summarised in Table 15)
provide a simple, concise way of distinguishing between the different implementation options.
The first column in the results table shows the list of TTC scheduler implementations which have
been tested in this study. The second four columns summarise the output task sequencing results
from all STCs in each TTC implementation. The sixth column then contains CPU overheads in

15
percentages. Columns 7 to 9 include: Tick jitter levels from STC C, the release jitter levels for
Task A and the release jitter levels for Task B, respectively, using STC B. It is worth noting that
these are the most appropriate measures for showing the impact on jitter as a result of using
various scheduler structures. Finally, the last three columns present the memory and power
overhead results for comparison purposes.
Table 15: Summary of results obtained in this paper.
Jitter in Task A has been included in the table to allow a comparison with the jitter levels in
lowpriority tasks. Key jitter results are shown in Figure 15 for comparison purposes. It can be
clearly noticed that the TTC-MTI and TTC-Adaptive schedulers had the ability to eliminate
release jitter in all tasks running in the system regardless of their order or position (see [30], [60]
for further details). The results for CPU, memory and power requirements are presented in Figure
16 to Figure 19 to facilitate a graphical comparison between all schedulers with respect to
hardware resource utilisation.
From the presented graphs, we can make the following observations. The CPU overheads in all
compared schedulers are approximately the same. However, such results have been presented in
the paper to demonstrate that the improvement achieved by some schedulers is not compromised
by their computational costs. On the other hand, it can be seen that the code memory required
to implement (for example) the TTC-MTI scheduler was even smaller than was used for the
majority of other schedulers. Particularly for the TTC-Adaptive scheduler, the little increase in the
code memory as compared to other schedulers is outweighed by the significant improvement
achieved in the scheduler predictability [60]. Moreover, it is so clear that the TTC-DVS has an
advantage over all other implementations in term of power requirement. This is due to the
incorporation of DVS technique which was intended to reduce the CPU power consumption in
TTC Scheduler [12].
Figure 15. Key jitter results in all TTC implementations

16
7. CONCLUSIONS
While there has been a great deal of interest in the development, assessment and refinement of
real-time scheduling algorithms, only limited amount of work has considered ways for bridging
the gap between scheduling algorithms and scheduler implementations in real-time,
resourceconstrained embedded systems. This is unfortunate because there is always a “one-to-
many” mapping between scheduler algorithms and practical scheduler implementations.
Therefore, it is expected that decisions made at the implementation stage will have a measurable
impact on the runtime behaviour of the embedded application.
In the study presented in this paper, we first sought to identify a set of representative
implementation classes for a TTC scheduler and then introduced the concept of “scheduler test
cases” (STCs) as a means of assessing the behaviour of these different scheduler forms. The aim
with these STCs was to facilitate empirical “black box” comparisons of the different scheduler
implementations and be able to distinguish the behaviour of such implementations without
necessitating access to (or understanding of) the underlying source code. Through the
employment of these STCs, we were able to fully document and compare the behaviour of TTC
implementations against some empirical measures: i.e. task behaviour (under normal conditions
and in the event of errors), tick and tasks release jitter, CPU and memory overheads, and finally
the CPU power consumption. Please note that, while we believe that STCs provide a useful way
of making quantitative empirical comparisons of different TTC scheduler implementations,
further work remains to be done in this area. For example, our current STCs consider only system
behaviour and do not take into account factors such as ease of use. It may be that empirical
techniques – such as the Small Group Methodology [67] – could be integrated with STCs in order
to take such factors into account. Despite this limitation, we still believe that the STC approach
has the great potential to apply on a wide range of scheduling algorithms other than TTC.

17
Certainly, if the complexity of an algorithm grows, the number of possible implementation
options of that algorithm would naturally grow. This would simply imply that the developed STC
technique is more required to verify (and possibly compare) the behaviour of the various
implementations of the tested algorithm. For example, alternative real-time scheduling algorithms
such as rate monotonic, earliest deadline first, least laxity first and priority ceiling protocol can be
considered as potential targets for the proposed STC technique. Since such algorithms are
generally more complicated than the TTC, designing (hence implementing) an appropriate set of
test cases would inevitably be a challenging process. Moreover, considering the application of
STC technique in multiprocessor embedded systems employing TTC scheduling algorithms
(along with shared-clock protocol for message scheduling) can be a good suggestion for future
publication.
ACKNOWLEDGEMENTS
Authors of this paper would like to thank Professor Michael Pont for supervising this project,
performed in the Embedded Systems Laboratory (ESL) research group, University of Leicester,
United Kingdom.
REFERENCES
[1] H. Kopetz, Real-Time Systems. Boston, Ma: Springer Us, 2011.
[2] M. Nahas, “Bridging The Gap Between Scheduling Algorithms And Scheduler Implementations In
Timetriggered Embedded Systems.,” Thesis, University Of Leicester, 2009.
[3] B. Sun, X. Li, B. Wan, C. Wang, X. Zhou, And X. Chen, “Definitions Of Predictability For Cyber
Physical Systems,” Journal Of Systems Architecture, Vol. 63, Pp. 48–60, Feb. 2016.
[4] M. J. Pont, Patterns For Time-Triggered Embedded Systems: Building Reliable Applications With
The 8051 Family Of Microcontrollers. Harlow: Addison-Wesley, 2001.
[5] M. Short, “Analysis And Redesign Of The ‘Ttc’ And ‘Tth’ Schedulers,” Journal Of Systems
Architecture, Vol. 58, No. 1, Pp. 38–47, Jan. 2012.
[6] M. Nahas, M. J. Pont, And M. Short, “Reducing Message-Length Variations In Resource-Constrained
Embedded Systems Implemented Using The Controller Area Network (Can) Protocol,” Journal Of
Systems Architecture, Vol. 55, No. 5–6, Pp. 344–354, May 2009.
[7] M. A. Hanif, “Design And Evaluation Of Flexible Time-Triggered Task Schedulers For Dynamic
Control Applications,” Thesis, Department Of Engineering, 2013.
[8] M. Nahas And A. M., “Ways For Implementing Highly-Predictable Embedded Systems Using
Timetriggered Co-Operative (Ttc) Architectures,” In Embedded Systems - Theory And Design
Methodology, K. Tanaka, Ed. Intech, 2012.
[9] T. P. Baker And A. Shaw, “The Cyclic Executive Model And Ada,” Real-Time Syst, Vol. 1, No. 1,
Pp.7–25, Jun. 1989.
[10] C. D. Locke, “Software Architecture For Hard Real-Time Applications: Cyclic Executives Vs. Fixed
Priority Executives,” The Journal Of Real-Time Systems, Vol. 4, No. 1, Pp. 37–53, Mar. 1992.
[11] G. Langelier, A. Dury, A. Petrenko, S. Ramesh, And T. Assaf, “Building An Interactive Test
Development Environment For Cyclic Executive Systems,” In Industrial Embedded Systems (Sies),
2015 10th Ieee International Symposium On, 2015, Pp. 1–9.
[12] T. Phatrapornnant And M. J. Pont, “Reducing Jitter In Embedded Systems Employing A Time-
Triggered Software Architecture And Dynamic Voltage Scaling,” Ieee Transactions On Computers,
Vol. 55, No. 2, Pp. 113–124, Feb. 2006.
[13] D. I. Katcher, H. Arakawa, And J. K. Strosnider, “Engineering And Analysis Of Fixed Priority
Schedulers,” Ieee Transactions On Software Engineering, Vol. 19, No. 9, Pp. 920–934, Sep. 1993.
[14] B. Koch, “The Theory Of Task Scheduling In Real-Time Systems: Compilation And Systematization
Of The Main Results,” Studies Thesis, University Of Hamburg, 1999.
[15] M. Nahas, “Studying The Impact Of Scheduler Implementation On Task Jitter In Real-Time
Resourceconstrained Embedded Systems,” Journal Of Embedded Systems, Vol. 2, No. 3, Pp. 39–52,
2014.

18
[16] M. W. Whalen And M. P. E. Heimdahl, “On The Requirements Of High-Integrity Code Generation,”
In 4th Ieee International Symposium On High-Assurance Systems Engineering, 1999.
Proceedings,1999, Pp. 217–224.
[17] E. Estévez, A. Sánchez-García, J. Gámez-García, J. Gómez-Ortega, And S. Satorres-Martínez, “A
Novel Model-Driven Approach To Support Development Cycle Of Robotic Systems,” Int J
AdvManufTechnol, Vol. 82, No. 1–4, Pp. 737–751, Jun. 2015.
[18] P. Marsh, “Models Of Control,” Electronics Systems And Software, Vol. 1, No. 6, Pp. 16–19,
Dec.2003.
[19] C. O’halloran, “Issues For The Automatic Generation Of Safety Critical Software,” In Ase, 2000, P.
277.
[20] B. Schätz, T. Hain, F. Houdek, W. Prenninger, M. Rappl, J. Romberg, O. Slotosch, M. Strecker, And
A. Wisspeintner, “Case Tools For Embedded Systems,” Tum-I, TuMünchen, 2003.
[21] K. Michels, “Trends In The Development Of Drive Components For Electric And Hybrid
Vehicles,”Atzelektronik Worldwide, Vol. 10, No. 4, Pp. 4–7, 2015.
[22] C. Mwelwa, K. Athaide, D. Mearns, M. J. Pont, And D. Ward, “Rapid Software Development For
Reliable Embedded Systems Using A Pattern-Based Code Generation Tool,” Sae Technical Paper,
2006.
[23] C. Mwelwa, M. J. Pont, And D. Ward, “Towards A Case Tool To Support The Development Of
Reliable Embedded Systems Using Design Patterns,” In Proceedings Of The 1st International
Workshop On Quality Of Service In Component-Based Software Engineering, 2003, Pp. 67–80.
[24] S. K. Baruah, “The Non-Preemptive Scheduling Of Periodic Tasks Upon Multiprocessors,” Real-
Time Systems, Vol. 32, No. 1–2, Pp. 9–20, Feb. 2006.
[25] T. Phatrapornnant, “Reducing Jitter In Embedded Systems Employing A Time-Triggered Software
Architecture And Dynamic Voltage Scaling,” Engineering, 2007.
[26] M. J. Pont, S. Kurian, H. Wang, And T. Phatrapornnant, “Selecting An Appropriate Scheduler For
Use With Time-Triggered Embedded Systems.,” In Proceedings Of The 12th European Conference
On Pattern Languages Of Programs (Europlop ’2007), Irsee, Germany, 2007, Pp. 595–618.
[27] C. Mwelwa, “Development And Assessment Of A Tool To Support Pattern-Based Code Generation
Of Time-Triggered (Tt) Embedded Systems,” Phd Thesis, University Of Leicester, Leicester,
Uk,2006.
[28] Y. Cho, S. Yoo, K. Choi, N.-E.Zergainoh, And A. A. Jerraya, “Scheduler Implementation In MpSoc
Design,” In Design Automation Conference, 2005. Proceedings Of The Asp-Dac 2005. Asia And
South Pacific, 2005, Vol. 1, Pp. 151–156 Vol. 1.
[29] S. Key, M. J. Pon, And S. Edwards, “Implementing Low-Cost Ttcs Systems Using Assembly
Language.,” In Proceedings Of The Eighth European Conference On Pattern Languages Of Programs
(Europlop 2003), Germany, 2003, Pp. 667–690.
[30] M. Nahas, “Employing Two ‘Sandwich Delay’ Mechanisms To Enhance Predictability Of Embedded
Systems Which Use Time-Triggered Co-Operative Architectures,” Journal Of Software Engineering
And Applications, Vol. 04, No. 07, Pp. 417–425, 2011.
[31] M. Nahas, “Implementation Of Highly-Predictable Time-Triggered Cooperative Scheduler Using
Simple Super Loop Architecture,” International Journal Of Electrical & Computer Sciences, Vol. 11,
Pp. 33–38, 2011.
[32] M. Nahas, M. J. Pont, And A. Jain, “Reducing Task Jitter In Shared-Clock Embedded Systems Using
Can,” In Proceedings Of The Uk Embedded Forum 2004, Birmingham, Uk, 2004, Pp. 184–194.
[33] M. Nahas, “Applying Eight-To-Eleven Modulation To Reduce Message-Length Variations In
Distributed Embedded Systems Using The Controller Area Network (Can) Protocol,” Canadian
Journal On Electrical And Electronics Engineering, Vol. 2, No. 7, Pp. 282–293, 2011.
[34] M. Nahas, “Developing A Novel Shared-Clock Scheduling Protocol For Highly-Predictable
Distributed Real-Time Embedded Systems,” American Journal Of Intelligent Systems, Vol. 2, No.
5,Pp. 118–128, Dec. 2012.
[35] Z. H. Hughes And M. J. Pont, “Design And Test Of A Task Guardian For Use In Ttcs Embedded
Systems,” In Proceedings Of The Uk Embedded Forum 2004, Birmingham, Uk, 2004, Pp. 16–25.
[36] Z. M. Hughes And M. J. Pont, “Reducing The Impact Of Task Overruns In Resource-Constrained
Embedded Systems In Which A Time-Triggered Software Architecture Is Employed,” Transactions
Of The Institute Of Measurement And Control, Vol. 30, No. 5, Pp. 427–450, Dec. 2008.
[37] R. L. Burdett And E. Kozan, “Techniques To Effectively Buffer Schedules In The Face Of
Uncertainties,”Computers& Industrial Engineering, Vol. 87, Pp. 16–29, 2015.
[38] F. Werner, “Scheduling Under Uncertainty,” Unpublished Document, 2012.

19
[39] I. Sommerville, Software Engineering. Addison-Wesley, 2007.
[40] E. Tran, “Verification/Validation/Certification,” Topics In Dependable Embedded Systems. Carnegie
Mellon University, 1999.
[41] Z. Liu, N. Gu, And G. Yang, “An Automate Test Case Generation Approach: Using Match
Technique,” In Computer And Information Technology, 2005. Cit 2005. The Fifth International
Conference On,2005, Pp. 922–926.
[42] P. Pringsulaka And J. Daengdej, “Coverall Algorithm For Test Case Reduction,” In Aerospace
Conference, 2006 Ieee, 2006, P. 8–Pp.
[43] H. Singh, M. Conrad, And S. Sadeghipour, “Test Case Design Based On Z And The Classification-
Tree Method,” In Formal Engineering Methods., 1997. Proceedings., First Ieee International
Conference On, 1997, Pp. 81–90.
[44] K. Tahera, C. Earl, And C. Eckert, “The Role Of Testing In The Engineering Product Development
Process,” 2012.
[45] T. Arts, J. Hughes, U. Norell, And H. Svensson, “Testing Autosar Software With Quickcheck,” In
Software Testing, Verification And Validation Workshops (Icstw), 2015 Ieee Eighth International
Conference On, 2015, Pp. 1–4.
[46] T. Kos, M. Mernik, And T. Kosar, “Test Automation Of A Measurement System Using A
Domainspecific Modelling Language,” Journal Of Systems And Software, Vol. 111, Pp. 74–88, 2016.
[47] S. J. Cunning And J. W. Rozenblit, “Automatic Test Case Generation From Requirements
Specifications For Real-Time Embedded Systems,” In Systems, Man, And Cybernetics, 1999. Ieee
Smc’99 Conference Proceedings. 1999 Ieee International Conference On, 1999, Vol. 5, Pp. 784–789.
[48] H. Do And G. Rothermel, “On The Use Of Mutation Faults In Empirical Assessments Of Test Case
Prioritization Techniques,” Software Engineering, Ieee Transactions On, Vol. 32, No. 9, Pp. 733–
752,2006.
[49] W. T. Tsai, L. Yu, X. X. Liu, A. Saimi, And Y. Xiao, “Scenario-Based Test Case Generation For
Statebased Embedded Systems,” In Performance, Computing, And Communications Conference,
2003.Conference Proceedings Of The 2003 Ieee International, 2003, Pp. 335–342.
[50] K. Beck, Extreme Programming Explained: Embrace Change. Addison-Wesley Professional, 2000.
[51] A. Hessel, “Model-Based Test Case Generation For Real-Time Systems,” 2007.
[52] A. J. Offutt And S. Liu, “Generating Test Data From Sofl Specifications,” Journal Of Systems And
Software, Vol. 49, No. 1, Pp. 49–62, 1999.
[53] A. Banerjee, S. Chattopadhyay, And A. Roychoudhury, “On Testing Embedded Software,” Advances
In Computers, 2016.
[54] M. M. Jaghoori, F. De Boer, D. Longuet, T. Chothia, And M. Sirjani, “Compositional Schedulability
Analysis Of Real-Time Actor-Based Systems,” ActaInformatica, Pp. 1–36, 2016.
[55] W. Hasanain, Y. Labiche, And S. Gheorghe, “Automated State-Based Online Testing Real-Time
Embedded Software With Rtedge,” In Model-Driven Engineering And Software Development
(Modelsward), 2015 3rd International Conference On, 2015, Pp. 294–302.
[56] M. Z. Iqbal, A. Arcuri, And L. Briand, “Environment Modeling And Simulation For Automated
Testing Of Soft Real-Time Embedded Software,” Software & Systems Modeling, Vol. 14, No. 1, Pp.
483–524,2015.
[57] J. Guan And J. Offutt, “A Model-Based Testing Technique For Component-Based Real-Time
Embedded Systems,” In Software Testing, Verification And Validation Workshops (Icstw), 2015 Ieee
Eighth International Conference On, 2015, Pp. 1–10.
[58] V. P. Kozyrev, “Estimation Of The Execution Time In Real-Time Systems,” Programming And
Computer Software, Vol. 42, No. 1, Pp. 41–48, 2016.
[59] M. J. Pont, S. Kurian, And R. Bautista-Quintero, “Meeting Real-Time Constraints Using ‘Sandwich
Delays,’” In Transactions On Pattern Languages Of Programming I, J. Noble And R. Johnson,
Eds.Springer Berlin Heidelberg, 2009, Pp. 94–102.
[60] M. Nahas And R. Bautista-Quintero, “Implementing Adaptive Time-Triggered Co-Operative
Scheduling Framework For Highly-Predictable Embedded Systems,” American Journal Of Embedded
Systems And Applications, Vol. 2, No. 4, Pp. 38–50, 2014.
[61] M. Nahas And A. Maait, “Choosing Appropriate Programming Language To Implement Software For
Real-Time Resource- Constrained Embedded Systems,” In Embedded Systems - Theory And Design
Methodology, K. Tanaka, Ed. Intech, 2012.
[62] “Lpc2100 Datasheet, Pdf - Alldatasheet.” [Online].
Available:Http://Www.Alldatasheet.Com/View.Jsp?Searchword=Lpc2100. [Accessed: 01-Jun-2016].
[63] I. Circuits, “Lpc2106/2105/2104 User Manual,” 2003.

20
[64] “Multifunction Devices-National Instruments.” [Online]. Available:
Http://Www.Ni.Com/Dataacquisition/Multifunction/.[Accessed: 01-Jun-2016]
[65] “Labview System Design Software - National Instruments.”
[Online].Available:Http://Www.Ni.Com/Labview/. [Accessed: 01-Jun-2016].
[66] G. C. Buttazzo, Hard Real-Time Computing Systems: Predictable Scheduling Algorithms And
Applications. New York: Springer, 2005.
[67] D. Ayavoo, “The Development Of Reliable X-By-Wire Systems: Assessingt He Effectiveness Of
A’simulationFirst’approach,” Engineering, 2006.
AUTHORS
MouaazNahas was born in UK on 1977. He received the B.Sc. degree (Electrical
Engineering) from Jordan University of Science and Technology, Jordan, in 2001, the
M.Sc. degree (Communications Engineering) from Loughborough University, UK, in
2002, and the Ph.D. degree (Embedded Systems) from University of Leicester, UK, in
2009. He is currently an Assistant Professor in the Department of Electrical Engineering
at Umm Al-Qura University, Makkah, Saudi Arabia. His main research interest is in the
development of cost-effective techniques for maximising the reliability of real-time,
resource-constrained embedded systems. He also has research interest in
electromagnetism and wireless communications.
Ricardo Bautista-Quintero was born in Mexico City on 1972. He received the B.Sc.
degree (Electronic Engineering) from InstitutoPolitécnico Nacional, Mexico, in 1995,
the M.Sc. degree (Microelectronics Engineering) from InstitutoPolitécnico Nacional,
Mexico in 2001, and the Ph.D. degree (Embedded Systems) from University of
Leicester, UK, in 2009. He is currently a Researcher Professor in the Department of
Mechanical Engineering at InstitutoTecnológico de Culiacan, Sinaloa, México. His
main research interest is in the development of reliable robust control implementations
on resource-constrained embedded systems.

DEVELOPING SCHEDULER TEST CASES TO VERIFY SCHEDULER IMPLEMENTATIONS IN TIMETRIGGERED EMBEDDED SYSTEMS

More Related Content

Similar to DEVELOPING SCHEDULER TEST CASES TO VERIFY SCHEDULER IMPLEMENTATIONS IN TIMETRIGGERED EMBEDDED SYSTEMS (20)

More from ijesajournal (20)

Recently uploaded (20)

DEVELOPING SCHEDULER TEST CASES TO VERIFY SCHEDULER IMPLEMENTATIONS IN TIMETRIGGERED EMBEDDED SYSTEMS