Incorporating Heterogeneity Principles in Embedded IOT Systems

Moore’s Law is not a law of nature, but an observation of a long-term trend in how technology is changing. With the help of Moore's law, which states that transistor density doubles every two years, technology is progressing rapidly. The Dennard Scaling, which states that power density would remain constant as transistor sizes decreased, is the complementary concept of Moore's Law. Developing software for embedded devices is a difficult task. Not only programs be resource efficient, as they operate under memory and timing constraints, but they should also take full advantage of the hardware to achieve maximum performance. In the heterogeneous design where many processing units like CPUs, GPUs, and FPGAs are merged into a single device was the result of this hunt for innovative solutions. However, with the high complexity of the systems and with the power and heat-dissipating constraints, multiple challenges are involved. This approach maximizes performance and efficiency, making it crucial for advancing technologies like AI, machine learning, and big data analytics. Advancing heterogeneous systems can help propel various sectors. It is essential to help advance heterogeneous embedded systems since it is such versatile technology. Heterogeneous computing is transforming how we handle complex computing tasks by combining different types of processors and specialized hardware. When designing such systems, performance and energy consumption are two important considerations. This study examines the possibilities and difficulties that a heterogeneous embedded system with embedded FPGAs and GPUs (as accelerators) might provide for applications, with an emphasis on performance and energy consumption. Throughout the research, we examine three design, modelling, and scheduling issues. This paper proposed the heterogeneity principles in the organizing and scheduling of network embedded/IoT systems.