Context and motivations. The desired behavior of a system in ubiquitous environments considers not only its correct functionality, but also the satisfaction of its non-functional properties, i.e., its quality of service. Given the heterogeneity and dynamism characterizing the ubiquitous environments and the need for continuous satisfaction of non-functional properties, self-adaptive solutions appear to be an appropriate approach to achieve interoperability. In this work, self-adaptation is adopted to enable software connectors to adapt the interaction protocols run by the connected components to let them communicate in a timely manner and with the required level of quality. However, this self-adaptation should be dependable, reliable and resilient to be adopted in dynamic, unpredictable environments with different sources of uncertainty. The majority of current approaches for the construction of self-adaptive software ignore the uncertainty underlying non-functional requirement verification and adaptation reasoning. Consequently, these approaches jeopardize system reliability and hinder the adoption of self-adaptive software in areas where dependability is of utmost importance. Objective. The main objective of this research is to properly handle the uncertainties in the non-functional requirement verification and the adaptation reasoning part of the self-adaptive feedback control loop of software connectors. This will enable a robust and runtime efficient adaptation in software connectors and make them reliable for usage in uncertain environments. Method. In the context of this thesis, a framework has been developed with the following functionalities: 1) Robust control of uncertainty in runtime requirement verification. The main activity in runtime verification is fine-tuning of the models that are adopted for runtime reasoning. The proposed stochastic approach is able to update the unknown parameters of the models at runtime even in the presence of incomplete and noisy observations. 2) Robust control of uncertainty in adaptation reasoning. A general methodology based on type-2 fuzzy logic has been introduced for the control of adaptation decision-making that adjusts the configuration of component connectors to the appropriate mode. The methodology enables a systematic development of fuzzy logic controllers that can derive the right mode for connectors even in the presence of measurement inaccuracy and adaptation policy conflicts. Results. The proposed model evolution mechanism is empirically evaluated, showing a significant precision of parameter estimation with an acceptable overhead at runtime. In addition, the fuzzy based controller, generated by the methodology, has been shown to be robust against uncertainties in the input data, efficient in terms of runtime overhead even in large-scale knowledge bases and stable in terms of control theory properties. We also demonstrate the applicability of the developed framework in a real-world domain. Thesis statement. We enable reliable and dependable self-adaptations of component connectors in unreliable environments with imperfect monitoring facilities and conflicting user opinions about adaptation policies by developing a framework which comprises: (a) mechanisms for robust model evolution, (b) a method for adaptation reasoning, and (c) tool support that allows an end-to-end application of the developed techniques in real-world domains.