With the development of communication in embedded systems, machine to machine communication becomes more and more important. The number of devices connected to the internet is growing exponentially, and intelligent devices are reaching more application areas than before. This leads to a fast growing complexity in all networked systems. Demands on manufacturing are also growing. The ability to dynamically adapt a manufacturing system to current needs has become a requirement to stay competitive. However, traditional systems are designed top-down, with each component selected to fulfill its specific purpose. This makes dynamic system reconfiguration a very challenging task that is proportional to the complexity of the system. The Arrowhead Framework is a proposed solution to manage this complexity by using a service oriented architecture. In this work a distributed synchronous measurement system within the ARROWHEAD framework, using the wireless 6LoWPAN MULLE platform developed for this framework, is implemented. While the ARROWHEAD framework focuses on communication between devices, the main challenge remaining for a distributed measurement system is clock synchronization, especially with low-cost Internet of Things devices. Therefore the focus of this work is to analyze the unique challenges that arise from software based clock synchronization, to develop mathematical error models and to propose methods to compensate these errors. The implementation of this work is based on the PTP protocol, however all the findings can be applied directly to all software based synchronization protocols. Software only timestamping leads to undetectable jitter from network conflicts and interference from other processes. Additionally the available timestamping points lead to transmission delay asymmetry. It will be shown that, by understanding these errors in detail, software based clock synchronization can be greatly improved.