Compiler-Assisted Thread Abstractions for Resource-Constrained Systems

Summary

Major operating systems for wireless sensor networks (WSN) enforce an event-based programming paradigm for efficiency reasons. However, practice has shown that the resulting code complexity is hard to manage for software developers and leads to difficult problems during development, deployment, and operations. Thus, thread libraries for WSN applications have been introduced, but the efficiency constraints of the domain often lead to glaring restrictions of the supported thread semantics.

In contrast, compiler-assisted thread abstractions support threads without actually using threads at runtime. Instead, the thread-based code is automatically translated into equivalent event-based code. Currently, two different systems implement this idea, but both have severe limitations regarding both the thread semantics and the tool support.

Our goal is to demonstrate the potential of compiler-assisted thread abstractions by taking the next step and introducing a comprehensive system of compiler and debugger which supports cooperative threads with minor restrictions and which is also platform independent.

A Haskell-based compiler prototype shows the feasibility of our approach and preliminary results demonstrate that the generated code is efficient enough to be executed on wireless sensor network devices. Furthermore, existing optimization potential suggests that compiler-assisted thread abstractions can indeed outperform runtime-based solutions and compete with hand-written event-based code. We are currently working on completing the implementation of the tools and verifying these points by means of an extensive evaluation.

A Comprehensive Compiler-Assisted Thread Abstraction for Resource-Constrained Systems

While size and complexity of sensor networks software has increased significantly in recent years, the hardware capabilities of sensor nodes have been remaining very constrained. The predominant event-based programming paradigm addresses these hardware constraints, but does not scale well with the growing software complexity, often leading to software that is hard-to-manage and error-prone. Thread abstractions could remedy this situation, but existing solutions in sensor networks either provide incomplete thread semantics or introduce a significant resource overhead. This reflects the common understanding that one has to trade expressiveness for efficiency and vice versa. Our work, however, shows that this trade-off is not inherent to resource-constrained systems. We propose a comprehensive compiler-assisted cooperative threading abstraction, where full-fledged thread-based C code is translated to efficient event-based C code that runs atop an event-based operating system such as Contiki or TinyOS. Our evaluation shows that our approach outperforms thread libraries and generates code that is almost as efficient as hand-written event-based code with overheads of 1% RAM, 2% CPU, and 3% ROM.

• A Comprehensive Compiler-Assisted Thread Abstraction for Resource-Constrained Systems
• Alexander Bernauer, Kay Römer
• Proceedings of the 12th ACM/IEEE Conference on Information Processing in Sensor Networks (IPSN 2013)

Programming Abstractions with Debugging Support for Resource-Constrained Devices

Abstractions are crucial in order to manage complex systems. In pervasive computing, though, common programming abstractions tend to be too expensive for the employed resource-constrained devices. In recent years, the wireless sensor network community has proposed several solutions to this problem. However, little has been done to also support debugging on the level of the abstraction. Instead, a developer is forced to understand the lower-level details in order to find and correct defects. This clearly hampers the development of applications. We aim at advancing the state of the art in programming of resource-constrained devices by introducing debugging support for programming abstractions.

Threads2Events: An Automatic Code Generation Approach

There is a long-standing dispute on whether and when thread-based programming should be preferred over the event-based paradigm. This dispute has also extended into the wireless sensor networks domain. Many existing operating systems rely on events due to their efficiency, but make code management difficult. Others rely on threads for developer comfort, but at the cost of reduced runtime efficiency. In this paper we try to combine the best of both worlds by offering a full-fledged cooperative thread abstraction with blocking I/O to the C programmer that is compiled into efficient event-based code. We present the basic code transformations and investigate their efficiency using a representative application case study. We find that RAM usage of generated code competes with hand-written code, but further optimizations are required to reduce the code size and the number of CPU cycles.

Meta-Debugging Pervasive Computers

As computers get more complex, the task of programming them gets more complex as well. This is especially true for the "Pervasive Computer", which is a massively distributed system consisting of unreliable embedded devices that communicate with each other over lousy wireless links. A common approach to address the programming problem is to offer programming abstractions that hide certain aspects of the complexity from the programmer. While several such abstractions and mappings thereof to low-level target languages have been proposed, there is a glaring lack of debugging support. It is typically impossible to debug at the conceptual level offered by the programming abstractions, instead one has to resort to debugging the generated target code. In this position paper we argue that programming abstractions should be designed in a way that allows debugging at the same conceptual level as programming. We further present requirements for such debugging tools, a taxonomy of programming abstractions and discuss debugging challenges, existing solutions, and potential approaches in each class.