Carriers and content providers are looking to boost revenue streams through the introduction of new applications and services such as data-rich and multimedia 3G services. Another key area of revenue growth for carriers is mobile workforce automation applications, which include tracking an organization's sales force or managing its field service using mobile technologies. Today's handset silicon designers, manufacturers, and software developers are building the next-generation mobile handsets that will provide the foundation for these new applications and revenue streams.
Given their large infrastructure investment to support enhanced multimedia and data services (such as broadcast video, instant messaging, and streaming music) and the need to ensure flawless deployment, carriers are increasingly demanding that handset manufacturers prove the reliability and consistency of their products over live networks before carriers are willing to release them to subscribers. This entails comprehensive testing that demonstrates not only functional capabilities but quality as well. To maximize their investment, handset manufacturers need automated tools and reusable test assets that carry not only between generations of a design but between different models and even different operating systems. In addition, since everyone in the value chain wants to guarantee that handsets reliably access services across live networks, being able to collaborate and share test assets leads to accelerated testing with higher reliability and less investment all around.
The Truth About Testing
An important part of testing handsets is being able to focus testing on these emerging services that promise to bring significant new revenue to carriers and content providers. For these services to be successful, they must be rolled out flawlessly. For example, early adopters who have trouble using consumer multimedia services will switch over to other carriers who have implemented these "have-to-have" services properly. For workforce automation applications, businesses will likewise drop services if they fail to reap expected productivity improvements.
Often these services are pitched as substantially simplifying operations and reducing costs, and subscribers have high expectations for them. Personal media players, for example, have already set a high quality expectation in consumer subscribers. For carriers, the stakes are high as this is the time when initial multimedia market share is being carved out. For businesses employing workforce automation, they are often attempting to mobilize critical business processes, which means they are basing the success of their enterprise on these services. In other words, they are counting on an error-free deployment.
There are three primary contributors in the handset value chain: the chipset vendors, operating system software developers, and handset manufacturers. These first three pretty much create the handsets that will be available, and everyone else in the value chain - application developers, content providers, carriers, and subscribers - must choose from and are limited to the choices they make available. While carriers don't directly contribute to the manufacturing of handsets, they are increasingly influencing handset specifications.
One fundamental consideration is that many of the tests performed across the value chain are, from a testing perspective, virtually identical. For example, both handset manufacturers and carriers need to verify functionality and quality of handsets within a carrier's live network. Certainly, if a carrier is going to invest the time to verify a particular handset, it wants to have reasonable confidence that the handset manufacturer has done all it can to make sure the handset will pass.
The reality is carriers want handsets to pass. At this time, however, carriers are concerned about revealing too much about what they test for fear that this would set a minimum baseline of what manufacturers would test. The key to resolving this issue is through automation with collaboration. In this way, carriers and handset manufacturers can work together to establish and develop handset verification tests and processes that streamline testing while meeting the verification requirements of all parties concerned.
Automating Test Processes
Many handset manufacturers and carriers rely on manual testing methodologies to verify handset functionality and quality, especially for multimedia services. For example, in order to verify the quality of a sports highlight video service, the quality of the received video must be tested in several ways, including image quality, playback start latency, and frames per second (fps), among others.
Testing video services manually is time-consuming, tedious, and prone to error. Technicians will make mistakes - either when entering commands or evaluating results - which can mask real problems. Manual key entry is a slow process and many tests are comprised of complex interactions. Manual testing requires additional personnel who must be retrained with each successive product generation or variation. Quality testing is subjective with results varying from hour-to-hour. In addition, many carriers manage test cases and test execution manually as well, using tools such as Microsoft Excel, which are not designed for the job.
In recent years, handset manufacturers and carriers have begun taking advantage of automated testing technology to simplify testing and manage the growing test burden. Automation increases testing consistency, conformance, and reliability. Testing platforms don't make typing mistakes and can operate 24/7 to accelerate testing. Errors cannot be overlooked and overall testing is faster because measurements and evaluations are computed quickly.
In addition, automation of testing processes enables development of reusable test assets. Rather than describe a test case specifically (i.e., which keys to press to send a text message), developers can utilize Adaptive Test Case technology to define test cases abstractly. Once created, these test assets can be used across different platforms, OSes, and even carrier networks (see Figure 1). They can also be shared across the value chain. Consider the value of being able to provide not only a handset but also a complete test suite verifying interoperability over a carrier's network. This simplifies testing for the carrier, accelerating the handset's adoption process. Furthermore, providing automated test assets ensures that carriers will test handsets properly as well as guarantee that manual testers will not accidentally fail handsets due to human error. Automated tests provide consistency between tests executed by handset manufacturers and carriers since interpretation plays no role in automated tests. If desired, carriers and manufacturers can also collaborate on testing guidelines, which can then be automated and added to the test assets as well.
Developing test assets that comprehensively verify the functionality of a handset is not a trivial undertaking. One of the major challenges for defining comprehensive tests, whether manual or automated, is determining all of the various modes of operation for a given handset function. For example, a subscriber can send a text message with just text, text with the current image just taken by the integrated camera, or even text with a saved image. In addition, this message could be sent to a new contact just typed in, to someone from the address book, in reply to a message, or broadcast to a group. The subscriber could have initiated this action from the main menu, by scrolling through various menus, or as an option from another application. When sending a message, the phone could also receive an incoming call, the subscriber could return to a game in progress, or another text message might be incoming.
What is being tested in all these cases is the robustness of the handset under different use-case scenarios. Carriers want to ensure that there isn't some hidden problem lurking in a handset that will cause the phone to malfunction or operate improperly when used in a certain way. However, in order to test each of these cases in this simple example, a developer would have to construct over 100 test cases (3 x 4 x 3 x 3). Fortunately, not every combination must be tested in order to obtain reasonably good test coverage. In most cases, testing the important and common use-cases as well as the various boundary conditions or operational extremes is a good start. Determining which test cases to implement and use is the most important part of defining the overall test bench. Even with pruning, however, it isn't uncommon for test suites to contain over a thousand test cases.
The reality is that developing an efficient handset verification process requires significant investment in the creation of test cases as well as a commitment of resources to perform the actual testing. Automation streamlines both the creation and execution of test processes, enabling the comprehensive test coverage required to ensure that handsets fail in the lab where errors can be corrected instead of in the field where carriers risk losing subscribers.
Connectivity Is Key to Automated Testing
In order for automated handset testing to work, the device under test must support some mechanism for being stimulated (i.e., having its keys virtually pressed in a way equivalent to a test technician manually pressing keys) and verifying a correct response.
Software-based connectivity mechanisms rely on software agents running on top of or within the operating system to provide test platforms with access to the user interface, internal handset resources, and sufficient capabilities for automating testing of common handset functions. Also software agents can transfer data, including screenshots, back to the test platform over standard device communication interfaces. These screenshots are used to verify proper handset behavior and measure handset performance, such as how long it took for a portal page to appear. With text and screen object recognition technology, a testing platform can even navigate menus dynamically based on context (i.e., find the option marked "Edit") rather than having to know a command's fixed position in a list.
Software agents are invaluable in providing access into handsets. Not only can handset manufacturers, content providers, and carriers test handset functionality, they can also measure the quality of services provided. For example, a testing platform could measure the average time to initiate a call or how quickly a Web page first appears (different from how long until it fully loads).
For multimedia services, testing extends to the perceived quality of images, video, and sound. Just because an image arrived doesn't mean it arrived intact or, as might be the case, that it is the expected image and displayed correctly. Using manual testing processes, human technicians must view the received multimedia data and provide a subjective assessment of its quality, typically using a quality rating scale such as Mean Opinion Score (MOS). Consistency in measurement, however, is difficult to achieve this way.
Automated testing tools now include objective methods for evaluating audio and video. Because of the high data rate associated with the capture of audio and video streams, hardware-based handset connectivity must be employed. Raw audio and video data are captured by the hardware interface and transferred to the testing platform in near real time. Quality assessment is performed in a PC test system where significant computer processing power exists to perform audio and video analysis without negatively impacting performance of the incident test.
Today, software agents can reasonably support the capture of about 4 fps from handsets. However, even with the relatively small amount of frames being captured, the function and behavior of the handset can be adversely impacted. To avoid this problem, hardware-based interfaces must be used. To make multimedia testing common place, handset manufacturers will need to provide hardware-based interfaces from the handset that are capable of passing audio and video data at 30 fps. In this way, a test platform can non-intrusively capture what the user sees on the handset screen and hears from the speaker without impacting the operation of the handset under test.
The Future of Multimedia Testing
It is essential that automated test connectivity become ubiquitous and commonplace. For this to happen, standard testing interfaces are needed. Two examples of existing test interfaces are Mobile Test Connectivity (MTC) from TestQuest and Spirent's Universal Tool Suite (UTS). MTC and UTS provide device stimulation and the screen capture capabilities required to evaluate image quality.
There is industry pressure from carriers for handset manufacturers to implement interfaces like MTC or UTS in handsets to standardize and simplify handset testing. In at least one case, the carrier tests the interface first and if it fails, rejects the handset without further testing.
Having standard interfaces is an important step in increasing the efficiency of the testing process. Such interfaces create an ecosystem around them of interoperable tools and products that facilitate sharing and collaboration throughout the value chain. MTC and UTS provide a test interface for basic handset functional testing. However, for multimedia testing, standard hardware interfaces are needed. Silicon vendors, handset manufacturers, and carriers need to work together to provide non-intrusive hardware access to data, audio, and video off the handset without involving software agents. Only in this way will it become possible to fully automate not only the functional testing of data and multimedia services but their quality as well.
Many handset manufacturers have already implemented a powerful test interface on their handsets but in most cases these interfaces are not available for use by carriers or application developers. Either the physical connector is removed or the software that supports them is disabled. Understandably, these manufacturers are concerned about the potential problems that could arise by leaving such interfaces exposed to hackers, although these concerns can be addressed by security technologies available today that can protect these interfaces and limit their use to legitimate testing.
In all likelihood, the adoption of a standard testing interface will be driven by the carriers, implemented by the silicon vendors, and supported at the operating system level. Manual testing of data and multimedia services is simply too difficult and costly, not to mention inadequate. A standard test interface that would enable automated testing of every functional aspect of a handset, as well as enable reliable quality testing of multimedia data, would provide tremendous value across the entire value chain, making even greater collaboration possible. Perhaps the most important consideration, however, is that providing a standardized test interface in handsets will yield a significant competitive edge to those handset manufacturers who open them up, because in the end the easier and more cost-effective manufacturers make it for carriers to verify handsets and developers to certify the applications that run on them, the better the chances of their handsets being adopted.
Most carriers recognize that mobile multimedia is still in its early days and they are proceeding with caution. Certainly, being first to market has its advantages but not if services fail to operate reliably. The potential revenues from large-scale deployments of new multimedia and automated workforce services are enticing, but for them to be successful, initial deployments must be flawless.
To ensure reliability, carriers need to be able to verify through comprehensive testing that handsets and services operate as expected over their networks. The complexity of manual testing requires that handset manufacturers, content providers, and carriers utilize automated testing technology to both reduce testing time and cost while increasing test coverage. Each group stands to gain much through automation and standardization of the testing process. By taking advantage of collaboration technologies, valuable test assets can be reused as well as shared across the entire value chain.
Together silicon designers, handset manufacturers, and software developers can work together to reduce overall testing complexity while increasing carrier confidence in a handset's ability to operate in a carrier's network. The end goal is to have carriers adopt handsets into their networks, and there are many ways in which the entire value chain can accelerate and improve the reliability of this process to the benefit of all.