Lock-Free Array Queues for 5G-NR Memory Deallocation

The patent divides the shared data structure into statically-sized regions organized as linked storage locations (e.g., lock-free array queues). Each region can be independently accessed and manipulated by different processes, allowing parallel operations without requiring global locking mechanisms. This segmentation enables processes to operate on different segments simultaneously, reducing the time spent waiting for memory operations to complete.

The patent implements lock-free data structures where each process independently manages its own access to shared data through atomic operations. Instead of requiring centralized locking mechanisms that slow down processing, each process performs self-service by using compare-and-swap (CAS) operations to ensure data consistency. This allows multiple processes to concurrently access and modify different parts of the shared data structure without blocking each other, maintaining data integrity while maximizing processing speed.

The patent pre-allocates statically-sized regions of memory for shared data structures before processes begin operations. This preliminary allocation eliminates the need for runtime memory allocation and deallocation operations, which are costly in terms of time. The pre-allocated regions are organized as fixed-size linked storage locations that can be efficiently accessed and manipulated without triggering additional memory management overhead, thus improving memory operation speed while maintaining sufficient flexibility for data sharing.