A Vacation-Based Performance Analysis of an Energy-Efficient Motorway Vehicular Communication System

Abstract

Due to the unprecedented growth in bandwidth requirement, the increasing number of access points (APs) deployed within a macrocell for services such as video conferencing, video gaming, and data off-loading leads to significantly higher energy consumption. This advancement in mobile networks has forced researchers to explore various methods of energy saving, although with little emphasis on motorway vehicular networks where mobility is also an important aspect. Energy saving in these networks is extremely challenging due to the dynamic nature of the environment in which they operate. To analyze such a network, we first develop a performance model for a medium access control (MAC) protocol, namely, the modified version of packet reservation multiple access (M-PRMA) with wireless channel impairments in a motorway vehicular environment. The M-PRMA protocol provides communication links (time slots) between an AP and the vehicles in range. The time slots of the M-PRMA protocol are modeled as servers where each outage of the channel is represented as a server on queue-length-independent vacation. Then, each AP, in a hierarchical micro-macro topology, is modeled as a single-server queue where the AP takes queue-length-dependent vacations (switches to sleep mode) to save energy during its inactivity period, although at the expense of degraded quality of service (QoS). To address this, a number of sleep strategies for the AP are studied. Finally, both of these proposed models (M-PRMA with channel impairments and AP with sleep cycles) are analyzed and verified through simulations. The performance results reveal that the introduction of sleep strategies at an AP can save up to 80% transmission energy during off-peak hours and 66% on average during the day in a motorway vehicular environment while supporting end-to-end QoS for video and audio conferencing applications.

Sleep-enabled roadside units for motorway vehicular networks

“…A number of studies showed that by considering a sleep strategy at a node during its inactivity, a certain amount of energy can be saved [2], [5], [9], [12], [13], [14], [15].…”

“…Hence, a macro-micro cellular structure is considered in this paper, where a number of micro-cells are served by the Access Points (APs) / Roadside Units (RSUs) within a macro-cell served by a BS [4], making it feasible to operate various sleep strategies for energy savings at the APs/RSUs. One such strategy is sleep cycles where an RSU switches OFF only its transmitter part for a randomly distributed time duration when there is no request to serve [9]. The RSU remains in such mode for the entire time duration (randomly generated with a certain mean value) even if packets are waiting to be served, thus degrading the system performance.…”

“…Moreover, saving energy through sleep cycles at the RSU incurs wake-up overheads, associated with each sleep cycle [9]. Therefore, there is a need to introduce various types of sleep cycles which not only improve the QoS but also reduce the wake-up overhead while maximising energy savings at the RSU in a motorway vehicular environment.…”

“…In [17], an M/M/c queue was analysed with queue length independent random vacations of the servers and their impact on average system delay and utilisation were studied. In [9], the authors studied two types of random vacations where the vacation duration was Negative exponential distributed using (1) queue length independent vacations to model the wireless channel impairments, and (2) queue length dependent vacations to model random sleep cycles (type-I). Both the arrival and the service discipline of packets were assumed to be memory-less Poisson type as a simplistic case.…”

“…The authors in [9] proposed a sleep strategy which reduced the power consumption of the RSU by switching OFF only its transmitter part for a randomly distributed time duration when there is no request to serve. The RSU remained in sleep mode for a fixed time duration (randomly generated with a certain mean value) even if packets were waiting to be served.…”