Modeling and Performance Analysis of Single-Server Database Over Quasi-static Rayleigh Fading Channel

Mengying Chen, Wannian An, Yang Liu, Chen Dong, Xiaodong Xu, Boxiao Han, Ping Zhang

Abstract

In wireless edge computing environments, database performance is inextricably linked to the underlying communication channel's quality. This paper proposes the Communication-Database (C-D) model, which treats the wireless channel as an integral part of the database service process. Using an M/G/1//K queuing model, the authors analyze how quasi-static Rayleigh fading affects transaction latency and system throughput. The results demonstrate that poor channel conditions lead to significant database congestion and increased waiting times, even when the database server itself is not overloaded.

Introduction

Traditional database research often overlooks the stochastic nature of the network, assuming constant transmission delays or infinite capacity. However, in Mobile Edge Computing (MEC), queries are sent over wireless channels prone to fading and noise. This paper addresses the disconnect between communication physics and database performance by modeling a single-server database over a quasi-static Rayleigh fading channel.

The core challenge is that fluctuating signal quality directly translates into variable service times from the user's perspective. When communication quality degrades, the "exclusive lock" held by a transaction persists longer, creating bottlenecks that ripple through the entire system.

The Communication-Database (C-D) Model

The researchers propose a tripartite model consisting of a finite source of $K$ clients, a communication system governed by Rayleigh fading, and a single-server database system utilizing exclusive locks.

The primary innovation is the definition of the total service time ($T_{all}$), which encompasses the uplink transmission ($T_{comup}$), the actual database processing time ($T_{serv}$), and the downlink transmission ($T_{comdown}$).

Timing diagram of the C-D model illustrating how communication delays contribute to the total service time $T_{all}$ and the impact of the timeout threshold $T$.

As shown in Figure 1, if the total time exceeds a predefined threshold $T$, the transaction is discarded. This model is analyzed using an M/G/1//K queue, where the service time distribution ($G$) is derived from the probability density functions of the Rayleigh fading channel. Two scenarios are evaluated: a nonlinear case (SISO) using Shannon capacity and a linear case (MIMO) where capacity scales linearly with power.

Performance Analysis and Results

The study relies on numerical simulations to validate theoretical derivations. The results show a near-perfect match between the closed-form mathematical solutions and simulated queuing behavior.

A key finding is the sensitivity of the system to the Rayleigh fading parameter ($\alpha$). As $\alpha$ increases—representing worsening channel conditions—the average waiting time ($W$) and queue length ($L$) grow exponentially.

Average waiting time $W$ versus arrival rate $1/\lambda$ for linear (L) and nonlinear (NL) cases across different fading parameters $\alpha$.

Figure 6 illustrates that for both linear and nonlinear models, higher fading parameters significantly shift the waiting time curves upward. Furthermore, the transition from SISO (nonlinear) to MIMO (linear) architectures provides a substantial reduction in the "staying time" of transactions, highlighting MIMO's role as a database performance optimizer.

Conclusion

The C-D model provides a robust framework for predicting database performance in wireless environments. It reveals that communication channels effectively act as part of the database's service mechanism. Future work will extend this model to multi-server (M/G/c//K) configurations and more complex fading scenarios, such as fast fading and shadowing, to better reflect high-mobility real-world applications.

Summary updated

BibTeX
@article{chen2023modeling,
  author = {Mengying Chen and Wannian An and Yang Liu and Chen Dong and Xiaodong Xu and Boxiao Han and Ping Zhang},
  title = {{Modeling and Performance Analysis of Single-Server Database Over Quasi-static Rayleigh Fading Channel}},
  journal = {IEEE Trans. Vehicular Technology},
  year = {2023},
  doi = {10.1109/tvt.2023.3247031},
}
Figure