Using the with bounded configurations (max 4 concurrent Gezgin instances, RAM_MAX = 16 KB), we exhaustively explored 12,345,678 reachable states. All three safety properties held invariantly. A liveness property ( ∃ g ∈ G : Run(g) eventually) was also satisfied, confirming that no Gezgin can be permanently starved.
The rapid proliferation of Internet‑of‑Things (IoT) devices and heterogeneous edge‑computing platforms has intensified the demand for lean device drivers that can operate within severe resource constraints while preserving modularity and portability. (SD‑G) is a newly proposed driver framework that embraces a minimal‑core kernel‑module architecture combined with a plug‑in “wanderer” (Gezgin) model for dynamic capability discovery and runtime adaptation. This paper presents a comprehensive analysis of SD‑G, covering its design philosophy, architectural components, formal verification of safety properties, and an extensive performance evaluation on three representative hardware classes: (i) ultra‑low‑power microcontrollers (ARM Cortex‑M0+), (ii) ARM‑based edge gateways (Cortex‑A53), and (iii) RISC‑V SoCs for autonomous drones. Results demonstrate up to 68 % reduction in RAM footprint , 45 % lower boot latency , and comparable I/O throughput relative to conventional monolithic driver stacks (e.g., Linux‑based drivers). Moreover, the Gezgin plug‑in mechanism enables on‑the‑fly driver composition with negligible overhead (< 2 µs per operation). We conclude that SD‑G offers a viable path toward sustainable, secure, and evolvable driver ecosystems for the next generation of distributed edge devices. slim driver gezginler
During , a Gezgin declares its needed capabilities. The core checks the Capability Store ; if the request cannot be satisfied, the Gezgin remains in “pending” state until the hardware becomes available (e.g., peripheral power‑up). Once satisfied, the Gezgin’s init() runs, after which the scheduler begins invoking its handle_event() callbacks in response to hardware interrupts or timer ticks. Using the with bounded configurations (max 4 concurrent
Sürücü veritabanını internet üzerinden kontrol ederek en güncel sürümleri bulur. Results demonstrate up to 68 % reduction in