PHDRDV / Institute for Rural Development / Over 50% of the world's economies face high enterprise failure with African countries being on the top of the list. Specifically, South Africa is among nations with the highest rate of enterprise failure, estimated at 70%, despite sustained investments by the government to support the growth of enterprises. It has been argued that the country has no entrepreneurship models to support enterprise development, as a result, policies for entrepreneurial-supporting initiatives are not driven by correct or informed mechanisms that can adequately transform enterprises. This study investigated the indicators and critical exogenous and endogenous components associated with successful enterprises, borrowing some aspects from the Economic base theory. A mixedmethod was followed and 280 participants were drawn from 16 villages in four local municipalities of Vhembe District using snowball, purposive and cluster sampling techniques. A desktop review, semi-structured and structured questionnaires were tools used for the data collection. An analysis of the qualitative data was achieved through a thematic technique using MAX QDA and Atlas-ti v8. Microsoft Excel functions; descriptive statistics through STATA, while, Crosstabulation, Principal Component Analysis (PCA) and Multilayer Perceptron (MLP) model through SPSS v26, as well as Multiple Linear Regression (MLR) model using R v3.0, were exploited with the quantitative data. The results indicate that - profit margin, trends of new products, enterprise expansion and enterprise survival - are common success indicators in the area. The PCA fitted on exogenous data structure (n=280) computed 6 principal enterprise challenges from 45 items identified qualitatively. These are - Access to finance (AF: 14.887%), Access to market (AM: 10.297%), Physical capacity (PC: 8.858), Operational cost (OC: 6.052%), Socio-cultural issues (SC: 5.628%) and Competition (Co 4.460%). The MLP based on 83 sample structure of success enterprises, however, revealed that Co presents the most challenge followed by AM, OC, SI, AF and PC which was the least challenging. Similarly, PCA post-endogenous qualitative study computed 5 principal components from 49 initial items. Bridging networks (BN) constituted 38.044% of the variance followed by Self-belief (SB:15.802), Risk Awareness (RA:6.144), Resilience (R: 4.532), and Nonconformist (NC:4.271). The MLR employed to investigate the linear relationship of the parameters revealed that BN (𝛽1 = 7.57) is most influential and statistically significant (p=0.01). Except for SI which is negatively related to enterprise success, R, RA and NC parameters demonstrate positive influences to enterprise success. A model for successful enterprises centred on entrepreneurs’ exogenous and endogenous attributes is proposed as the main contribution of the study towards enterprises’ success in the areas of the research. The key recommendation in this study is that support to enterprises should be informed by the area-specific indicators outlined in the study. Keywords: African model, enterprise development, endogenous attributes, exogenous factors, rural areas / NRF

Information and communication systems enter an increasing number of areas of daily lives. Our reliance and dependence on the functioning of such systems is rapidly growing together with the costs and the impact of system failures. At the same time the complexity of hardware and software systems extends to new limits as modern hardware architectures become more and more parallel, dynamic and heterogenous. These trends demand for a closer integration of formal methods and system engineering to show the correctness of complex systems within the design phase of large projects. The goal of this thesis is to introduce a formal holistic approach for modeling, analysis and synthesis of parallel systems that potentially addresses complex system behavior at any layer of the hardware/software stack. Due to the complexity of modern hardware and software systems, we aim to have a hierarchical modeling framework that allows to specify the behavior of a parallel system at various levels of abstraction and that facilitates designing complex systems in an iterative refinement procedure, in which more detailed behavior is added successively to the system description. In this context, the major challenge is to provide modeling formalisms that are expressive enough to address all of the above issues and are at the same time amenable to the application of formal methods for proving that the system behavior conforms to its specification. In particular, we are interested in specification formalisms that allow to apply formal verification techniques such that the underlying model checking problems are still decidable within reasonable time and space bounds. The presented work relies on an exogenous modeling approach that allows a clear separation of coordination and computation and provides an operational semantic model where formal methods such as model checking are well suited and applicable. The channel-based exogenous coordination language Reo is used as modeling formalism as it supports hierarchical modeling in an iterative top-down refinement procedure. It facilitates reusability, exchangeability, and heterogeneity of components and forms the basis to apply formal verification methods. At the same time Reo has a clear formal semantics based on automata, which serve as foundation to apply formal methods such as model checking. In this thesis new modeling languages are presented that allow specifying complex systems in terms of Reo and automata models which yield the basis for a holistic approach on modeling, verification and synthesis of parallel systems. The second main contribution of this thesis are tailored branching-time and alternating time temporal logics as well as corresponding model checking algorithms. The thesis includes results on the theoretical complexity of the underlying model checking problems as well as practical results. For the latter the presented approach has been implemented in the symbolic verification tool set Vereofy. The implementation within Vereofy and evaluation of the branching-time and alternating-time model checker is the third main contribution of this thesis.

Information and communication systems enter an increasing number of areas of daily lives. Our reliance and dependence on the functioning of such systems is rapidly growing together with the costs and the impact of system failures. At the same time the complexity of hardware and software systems extends to new limits as modern hardware architectures become more and more parallel, dynamic and heterogenous. These trends demand for a closer integration of formal methods and system engineering to show the correctness of complex systems within the design phase of large projects. The goal of this thesis is to introduce a formal holistic approach for modeling, analysis and synthesis of parallel systems that potentially addresses complex system behavior at any layer of the hardware/software stack. Due to the complexity of modern hardware and software systems, we aim to have a hierarchical modeling framework that allows to specify the behavior of a parallel system at various levels of abstraction and that facilitates designing complex systems in an iterative refinement procedure, in which more detailed behavior is added successively to the system description. In this context, the major challenge is to provide modeling formalisms that are expressive enough to address all of the above issues and are at the same time amenable to the application of formal methods for proving that the system behavior conforms to its specification. In particular, we are interested in specification formalisms that allow to apply formal verification techniques such that the underlying model checking problems are still decidable within reasonable time and space bounds. The presented work relies on an exogenous modeling approach that allows a clear separation of coordination and computation and provides an operational semantic model where formal methods such as model checking are well suited and applicable. The channel-based exogenous coordination language Reo is used as modeling formalism as it supports hierarchical modeling in an iterative top-down refinement procedure. It facilitates reusability, exchangeability, and heterogeneity of components and forms the basis to apply formal verification methods. At the same time Reo has a clear formal semantics based on automata, which serve as foundation to apply formal methods such as model checking. In this thesis new modeling languages are presented that allow specifying complex systems in terms of Reo and automata models which yield the basis for a holistic approach on modeling, verification and synthesis of parallel systems. The second main contribution of this thesis are tailored branching-time and alternating time temporal logics as well as corresponding model checking algorithms. The thesis includes results on the theoretical complexity of the underlying model checking problems as well as practical results. For the latter the presented approach has been implemented in the symbolic verification tool set Vereofy. The implementation within Vereofy and evaluation of the branching-time and alternating-time model checker is the third main contribution of this thesis.