SD
S.H.W. Dukker
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
2 records found
1
Static Kubernetes manifest validation mainly focuses on syntax and best practices. Many deployment failures are caused by syntactically valid resource definitions that prevent communication between dependent micro-services. This thesis investigates to how static analysis can detect such internal routing misconfigurations in Kubernetes manifests. Moreover, it also investigates the impact of misconfiguration detection by adding dependency context to the analysis. First, the Kubernetes resource model is reduced to a minimal set of resources and attributes that are required to reason about internal reachability. Next, runtime reachability issues are identified that are accepted by Kubernetes and remain undetected by existing validation tools. Based on these findings, the prototype static analysis tool KubeReach is developed, followed by KubeReach+, which extends the analysis with explicit dependency context of the micro-service architecture. The results show that Kubernetes manifests already contain sufficient information to detect several internal routing misconfigurations, but not enough to verify whether the the micro-services can reach its dependencies. Adding dependency context enables the validation of expected communication paths and improves the detection of routing issues that cannot be detected from the Kubernetes manifest alone. This makes dependency-aware static analysis a valuable addition to Kubernetes manifest integritiy.
...
Static Kubernetes manifest validation mainly focuses on syntax and best practices. Many deployment failures are caused by syntactically valid resource definitions that prevent communication between dependent micro-services. This thesis investigates to how static analysis can detect such internal routing misconfigurations in Kubernetes manifests. Moreover, it also investigates the impact of misconfiguration detection by adding dependency context to the analysis. First, the Kubernetes resource model is reduced to a minimal set of resources and attributes that are required to reason about internal reachability. Next, runtime reachability issues are identified that are accepted by Kubernetes and remain undetected by existing validation tools. Based on these findings, the prototype static analysis tool KubeReach is developed, followed by KubeReach+, which extends the analysis with explicit dependency context of the micro-service architecture. The results show that Kubernetes manifests already contain sufficient information to detect several internal routing misconfigurations, but not enough to verify whether the the micro-services can reach its dependencies. Adding dependency context enables the validation of expected communication paths and improves the detection of routing issues that cannot be detected from the Kubernetes manifest alone. This makes dependency-aware static analysis a valuable addition to Kubernetes manifest integritiy.
Localization is an important element for a moving swarm of robots. A swarm contains many individuals, and it is essential that the swarm members do not collide. Expensive and complex implementations to localize others is undesirable. Hence, a localization system designed using the available resources is desired. The available techniques that have been analysed are Received Signal Strength Indicator (RSSI), Time difference of arrival (TDOA) using the ESP-NOW protocol and a spinning time of flight (TOF) sensor. The spinning TOF sensor showed to be the most promising, with close range distance detection only containing a maximum errors of 5cm. The sensor is implemented as a cheap LiDAR system by mounting it to the front of the robot, which spins around its own axis. An IMU is responsible for keeping track of the orientation of the TOF sensor. Furthermore, two small algorithms were designed and compared in order to process the TOF data.
...
...
Localization is an important element for a moving swarm of robots. A swarm contains many individuals, and it is essential that the swarm members do not collide. Expensive and complex implementations to localize others is undesirable. Hence, a localization system designed using the available resources is desired. The available techniques that have been analysed are Received Signal Strength Indicator (RSSI), Time difference of arrival (TDOA) using the ESP-NOW protocol and a spinning time of flight (TOF) sensor. The spinning TOF sensor showed to be the most promising, with close range distance detection only containing a maximum errors of 5cm. The sensor is implemented as a cheap LiDAR system by mounting it to the front of the robot, which spins around its own axis. An IMU is responsible for keeping track of the orientation of the TOF sensor. Furthermore, two small algorithms were designed and compared in order to process the TOF data.