Mechelle Gittens

The vital few versus the trivial many: examining the Pareto principle for software

This paper discusses the Pareto principle as it relates to the distribution of software defects in code. We look at evidence in the context of both the software test team, and users of the software. We also investigate two related principles. The first principle is that the distribution of defects in code relates to the distribution of complexity in code. The second principle is that how we define complexity relates to the distribution of defects in code. We present this work as an empirical study of three general hypotheses investigated for large production-level software; we show that the essence of the principle holds, while precise percentages do not.

All code coverage is not created equal: a case study in prioritized code coverage

The increasing size and complexity of software systems has led to increased challenges in evaluating code coverage. In parallel, development projects for large complex software are increasingly limited in schedule and resources. Common code coverage analysis provides a complete measurement of testing breadth, but does not give guidance for coverage in a project setting. We propose a prioritized coverage approach that is indigenous to the development organization and provides capabilities for evaluating code coverage and setting priorities for testing. This paper presents an approach to code coverage analysis for large software that considers modern project constraints.

An iterative, multi-level, and scalable approach to comparing execution traces

In this paper, we overview a new approach to comparing execution traces. Such comparison can be useful for purposes such as improving test coverage and profiling systems users. In our approach, traces are compressed into different levels of compaction and are then compared iteratively from highest to lowest levels, rejecting dissimilar traces in the process and eventually leaving residual, similar traces. These residual traces form an important feedback for improvement or analysis goals. The preliminary results show that the approach is scalable for industrial use.

An empirical study on the use of mutant traces for diagnosis of faults in deployed systems

Debugging deployed systems is an arduous and time consuming task. It is often difficult to generate traces from deployed systems due to the disturbance and overhead that trace collection may cause on a system in operation. Many organizations also do not keep historical traces of failures. On the other hand earlier techniques focusing on fault diagnosis in deployed systems require a collection of passing-failing traces, in-house reproduction of faults or a historical collection of failed traces. In this paper, we investigate an alternative solution. We investigate how artificial faults, generated using software mutation in test environment, can be used to diagnose actual faults in deployed software systems. The use of traces of artificial faults can provide relief when it is not feasible to collect different kinds of traces from deployed systems. Using artificial and actual faults we also investigate the similarity of function call traces of different faults in functions. To achieve our goal, we use decision trees to build a model of traces generated from mutants and test it on faulty traces generated from actual programs. The application of our approach to various real world programs shows that mutants can indeed be used to diagnose faulty functions in the original code with approximately 60-100% accuracy on reviewing 10% or less of the code; whereas, contemporary techniques using pass-fail traces show poor results in the context of software maintenance. Our results also show that different faults in closely related functions occur with similar function call traces. The use of mutation in fault diagnosis shows promising results but the experiments also show the challenges related to using mutants.

Post-diagnosis management of diabetes through a mobile health consultation application

Diabetes currently ranks among the highest threats to human life given the increase in the number of diagnosed cases worldwide. This sudden increase has been linked to changes in human lifestyle since the majority of cases diagnosed are that of type 2 diabetes. Mobile health (m-health) technologies are being implemented in all areas of the health industry to aid patients in their pursuit of healthier lives. The society chosen for our study has a population predominately of African descent and is in crisis, as it possesses one of the highest rates of diabetes and amputation worldwide. Work was therefore conducted to promote proactive behaviour among members of the society by placing relevant technology in their hands to take control of their health. The purpose was to determine if the development of an m-health application would influence people diagnosed with diabetes to adopt healthier lifestyles, and reduce the impact of the disease on this part of the African diaspora. A mobile application was developed based on a preliminary study of m-health application availability and peoples attitudes towards mobile applications. People who had diabetes or caretakers of people with diabetes tested the mobile application over a two-week period. Average participation among all users was low but notably there were a few persons who used the application frequently. The frequent users were compliant with the expected behaviours of a diabetes patient. The other participants used the application infrequently and generally had bad health regimes.

A Real-time Application to Predict and Notify Students about the Present and Future Availability of Workspaces on a University Campus

The many responsibilities of students require that they access and utilize information about workspaces to facilitate time management. Workspaces such as computer labs, classrooms, and study areas have varying availability. At the University of the West Indies - Cave Hill Campus (UWICHC), students do not have a means of knowing in advance if these areas are unavailable other than to visit the space. If the space is unavailable, the student would have lost time. In addition, an initial survey of the opinions of 100 UWICHC students showed that 72% of the respondents found when a non-class participant entered a room where they were having a class, the activity distracted either them or the instructor. In this paper, we present a system which assesses the availability status of the room and makes a real-time occupancy indicator available to students on and off campus. This indicator avoids cases where a non-class participant interrupts an ongoing class. Additionally, students can query the system for a prediction or estimation on what the status of the space will be on a given day and time based on past conditions. In our survey, 91% of the respondents believed that having a system that relayed real-time or predictive information about class availability would help them, and therefore allow more time for productive activities. We also present an analysis of the use of the system and how students perceived the usefulness of the system in operation.

Identifying Recurring Faulty Functions in Field Traces of a Large Industrial Software System

Software maintainers use the traces of field failures to understand and diagnose faulty functions that cause the system to fail. Despite their usefulness, traces from the field can be quite overwhelming, especially for software systems with a vast client base. In the execution of realistic applications, many of them being millions of lines of code, there are just too many traces that are generated. In addition, traces are known to be extraordinarily large, which further complicates matters. Fortunately, not all field failures are caused by new faults. In fact, previous studies showed that 50% to 90% of field failures are due to previously known faults. In this paper, we propose a machine learning approach that automatically detects recurring faulty functions in the traces of new field failures. We achieve our goal by training decision trees on earlier resolved traces of system failures from the current and prior releases of the system. When applied to a large industrial system with 20 million lines of code and 200,000 functions, our approach was able to detect recurring faulty functions in the traces of field failures with an accuracy of 90%, to even 97% in some cases.

Focused iterative testing: a test automation case study

Timing-related defects are among the most difficult types of defects to catch while testing software. They are by definition difficult to reproduce and hence difficult to debug. Not all components of a software system have timing-related defects. For example, either a parser can analyze an input or it cannot. However, systems that have concurrent threads such as database systems are prone to timing-related defects. As a result, software developers must tailor testing to exploit vulnerabilities that occur because of threading. This paper presents the Focused Iterative Testing (FIT) approach, which uses a repetitive and iterative approach to find timing-related defects and target product areas with multithreaded characteristics by executing system tests with a multi-user test suite.

Monitoring and Analyzing Wi-Fi Availability and Performance on a University Campus Using Recycled Cell Phones to Aid Students in Selecting Study Areas

Wi-Fi availability on university campuses has become increasingly important since services such as email and assignment submission portals for students require them to have internet access on and off campus. This is the case at the University of the West Indies - Cave Hill Campus (UWICHC) where students are required to conduct several online activities including submitting assignments, checking grades and registering for classes online. A survey of opinions of 115 UWICHC students indicated that 96% of the respondents use Wi-Fi to access student services offered by the campus. Ninety-four (94%) also indicated they would like to have access to Wi-Fi information on active hotspots and signal strength in various study areas. In this paper, we present a green system that uses recycled cell phones to collect information about the Wi-Fi status in study areas. The information includes Wi-Fi signal strength, connection speed, download speed and internet availability and is shown to students and staff using a web application. The web/mobile application also displays information on access points in various study areas on a geographical map of the campus. Additionally, the application analyzes the data collected and gives students and IT staff peak times when students encounter difficulty such as poor connectivity or no connection to the internet suggesting that an access point is damaged or needs to be reset. Such information is also beneficial to the IT Department and can increase their response time at the low costs gained from using recycled cell phones sensors.

Towards an emerging theory for the diagnosis of faulty functions in function-call traces

When a fault occurs in the field, developers usually collect failure reports that contain function-call traces to uncover the root causes. Fault diagnosis in failure traces is an arduous task due to the volume and size of typical traces. Previously, we have conducted several research studies to diagnose faulty functions in function-call level traces of field failures. During our studies, we have found that different faults in closely related functions occur with similar function-call traces. We also infer from existing studies (including our previous work) that a classification or clustering algorithm can be trained on the function-call traces of a fault in a function and then be used to diagnose different faults in the traces where the same function appears. In this paper, we propose an emerging descriptive theory based on the propositions grounded in these empirical findings. There is scarcity of theorizing empirical findings in software engineering research and our work is a step towards filling this gap. The emerging theory is stated as: a fault in a function can be diagnosed from a function-call trace if the traces of the same or a different fault in that function are already known to a clustering or classification algorithm. We evaluate this theory using the criteria described in the literature. We believe that this emerging theory can help reduce the time spent in diagnosing the origin of faults in field traces.