4.2 Related Work and Contribution
4.2.1 Selective Literature Survey
The authors of [28] indicated that the DTI application of V2I communications suffered from a random access problem. Indeed, a close observation of a DTI sys- tem reveals the highly probable existence of multiple vehicles within the range of a RSU. This, together with the fact that more than one of these vehicles may simul- taneously require Internet access gives rise to a joint random access and spectrum
allocation problem whose resolution is challenging. To this end, the authors devel- oped the Dynamic Optimal Random Access (DORA) algorithm with the objective of maximizing the channel utilization subject to time-varying contention severity and capacity levels. DORA was developed in the context of a DTI scenario consisting of vehicles communicating with a single RSU. It incorporated self-incurred channel access fees and accounted for the Quality-of-Service (QoS) levels required by differ- ent applications. A finite-horizon Markov Decision Process (MDP)-based model was formulated to capture the system’s dynamics as well as to evaluate its performance when operating under DORA.
In a similar scenario to [28], the work of [29] examined the V2I wireless access for streaming applications in a public transportation system. The authors formulated an optimization problem with the objective of providing a cost-minimal wireless connec- tivity that satisfies the end-users QoS requirements. For this purpose, a hierarchical optimization framework was established to determine an optimal policy indicating whether or not it is expedient for a vehicle to request bandwidth reservation from the RSU. The proposed mathematical model studied the system’s performance variation as a function of the following factors: a) the streaming application’s requirements,
b) the vehicular mobility and c) the channel quality. The work in [29] considered
both the user-centric and network-centric point of views to provide a unified model for optimizing wireless access in a V2I communication scenario.
The authors of [30] modelled the vehicular data download process using a series of transient Markov Reward Processes. Their objective was to characterize the dis- tribution of a vehicle’s downloaded data volume throughout its residence time within a RSU’s range. The authors computed the influence of traffic density, vehicle speed, and RSUs transmission range on the amount of downloaded data.
where the RSU stored the Service Requests (SRs) and the request with the leastD∗S
was served first. D is the SR’s deadline and S is the data size to be uploaded to the
RSU.D∗S showed better performance when compared to three other access schemes
namely a) FCFS, b) Earliest Deadline First, and c) Smallest Datasize First. The authors then worked on improving the performance of their proposed algorithm by using broadcasting techniques, and hence, serving more requests simultaneously. Fur- thermore, in an attempt to study the tradeoffs between the service ratio and the data quality, the authors extend their algorithm and propose a Two-Step scheme where two priority queues were used, i.e., one for upload requests and the other for down- load requests. The results presented therein showed that the Two-Step scheduling scheme is adaptive to different workload scenarios. The authors then studied the uplink MAC performance of a DTI scenario in [32]. Both the contention nature of the uplink and the realistic traffic model were taken into consideration. An analyti- cal framework was developed to quantify the uplink performance of DTI in an IEEE 802.11p environment in terms of packet collisions and uplink capacity. Furthermore, for the purpose of maintaining optimal system performance, the authors explored the adjustment of transmission power as a means of admission control by the roadside unit.
The work of [62] revolved around modelling the RSU as a multi-server queue whose customers are Service Requests (SRs) generated by newly incoming vehicles into the RSU’s range. SRs will queue into the RSU’s buffer until either they get served or they renege from the queue due to the departure of their initiating vehicles from the RSU’s range. Upon the departure of a vehicle from the RSU’s range, all of its associated SRs will be either discarded if they are still queueing in the RSU’s buffer or force-terminated in case they are being served. The proposed complex model in [62] was partially simplified through approximations.
The authors of [63] aimed to improve the video quality and reduce its playback delay in a DTI network scenario. For this purpose, the authors propose a Selective Downlink Scheduling (SDS) algorithm whose main objective is to exploit information on the vehicles’ positions in order to maximize the amount of data each vehicle receives before leaving the roadside unit’s coverage. Their priority-based SDS algorithm is deployed at the roadside unit to coordinate the transmission of packets according to their importance, playback deadline as well as real-time information of vehicles such as the velocity, link quality and residence time within the range of the roadside unit. Finally, To guarantee near-absolute service differentiation in VANETs supporting multimedia applications with different QoS requirements, a control-theoretic packet scheduling algorithm was proposed in [27]. This algorithm relies on the polling-based contention-free access method of the IEEE 802.11e standard and adapts resource allocation to many factors such as queue length and vehicle residence time.