Abstract
Modern wireless cellular networks (WCNs) carry different types of services, including voice, video, data, web traffic, etc. In order to provide high-quality multimedia services to anyone, anywhere, at any time with the consideration of user mobility, network designers need to implement optimal mechanisms that can trade off between Quality of Service (QoS) and Channel Utilization (in other words, Resource Utilization). This paper presents a dynamic guard channel allocation (DGCA) scheme, which can adaptively adjust the number of reserved channels for handoff calls to achieve this goal. Heavy-tailed distributions are used to represent the multimedia traffics. SimPy is used to build the simulation model.
Introduction
In wireless cellular networks, user experience is primarily impacted by two factors: New Call Blocking and Handoff Call Dropping. While blocking a new call is inconvenient, dropping an ongoing call as a user moves between cells is significantly more disruptive. Consequently, network designers prioritize minimizing handoff drops over preventing new connections.
A common solution is the use of "Guard Channels" (GC), where a fixed number of channels are reserved exclusively for handoff calls. However, static reservation is often inefficient; it can lead to unnecessary blocking of new calls during low handoff demand or insufficient protection during traffic spikes. This paper addresses the need for a more flexible, self-adaptive approach that accounts for the diverse traffic patterns of modern multimedia services.
Methodology and Traffic Modeling
The authors propose a self-adaptive Dynamic Guard Channel Allocation (DGCA) scheme. This scheme monitors the Handoff Dropping Probability ($P_{DH}$) in real-time. If $P_{DH}$ falls below a minimum threshold ($Thr_1 = 0.5$), the system reduces the number of reserved channels ($g$) to accommodate more new calls. Conversely, if $P_{DH}$ exceeds a maximum threshold ($Thr_2 = 0.8$), $g$ is increased (up to a maximum of 12) to prioritize ongoing connections.
To ensure realistic results, the simulation moves beyond simple Poisson models. Modern multimedia traffic is "bursty," characterized by long-range dependence. The study utilizes heavy-tailed distributions, such as Pareto and Weibull, to model arrival processes and call durations for different service types.
The simulation was implemented using SimPy, a Python-based discrete-event framework, modeling a network cell with a 60-channel capacity over 10,000 simulation time units.
Results and Discussion
The performance of the DGCA scheme (Scheme 2) was compared against a traditional fixed-GC model (Scheme 1). The primary metric used was the Grade of Service (GoS), which weights handoff drops ten times more heavily than new call blocks ($W=10$), reflecting the industry's prioritization of connection continuity.
As shown in the simulation results, Scheme 2 consistently achieved a lower GoS across all traffic types—voice, data, video, and web—as the handoff call arrival rate increased. This indicates a superior balance between service quality and resource availability. Furthermore, the DGCA scheme demonstrated generally higher Channel Utilization (CU) by dynamically releasing reserved channels when they were not strictly required for handoffs, allowing the network to support a higher total volume of traffic.
Conclusion
The DGCA scheme provides a robust, closed-loop control mechanism for resource allocation in WCNs. By adapting to real-time network conditions and accounting for the bursty nature of multimedia traffic through realistic statistical modeling, it outperforms static reservation methods in both service quality and efficiency.
Future research could explore the sensitivity of the chosen thresholds ($0.5$ and $0.8$) and evaluate the scheme's performance under more complex mobility patterns, such as high-speed transit environments. Additionally, while the logic is computationally efficient, testing on physical base station hardware would validate its real-world feasibility.
Summary updated