• A multi-objective optimization model for throughput-delay trade-off in data stream parallel sharing. • High-dimensional state space modeling for dynamic delay-network relationship. • Adaptive weight coefficient balances throughput and delay based on real-time network state. • Three-layer data flow architecture enables efficient parallel sharing and processing. • Achieves maximum throughput and effectively reduces data stream transmission delay. Aiming at the problems of insufficient throughput and high latency in parallel sharing of data streams in the "dual carbon" digital intelligent monitoring center, a parallel sharing optimization method that balances throughput and latency is proposed. Firstly, an objective function is established with the goal of maximizing throughput, and the optimal throughput optimization strategy is solved under the constraints of link bandwidth capacity and load balancing. Secondly, to quantify the dynamic relationship between data stream transmission delay and network state, the concept of high-dimensional state space is introduced, and time-varying parameters such as link utilization, buffer occupancy, and channel quality are constructed as state vectors. Based on this, the transmission delay encoding function is derived. On this basis, a second objective function is established with the goal of minimizing network latency, and it is weighted and fused with the throughput objective function through weight coefficients to construct a multi-objective joint optimization model that considers both throughput and latency, achieving comprehensive optimization of data stream transmission performance. Furthermore, the data flow is divided into three layers: data collection and preprocessing layer, data storage and management layer, and data analysis and decision support layer. A data flow access sharing architecture is constructed to achieve parallel sharing optimization of data flow in the "dual carbon" digital intelligent monitoring center. The experimental results show that this method has the maximum throughput at different cycles and can effectively reduce the delay of data stream parallel sharing optimization, with good sharing optimization effect.
Sha et al. (Fri,) studied this question.