Service computing, as a bridge between modern business services and information technologies, has witnessed the communication infrastructure and environment, which the service-oriented architecture (SOA) build on, varied from wired network to wireless mobile network, from centralized client-server model to distributed peer-to-peer (p2p) model until today’s centralized cloud data centers in the past decade. The quality of service (QoS) is always one of the main spots that the users and researchers concern with in both service computing and mobile cloud computing area [
1‐
6]. A service provider should ensure the consistency of the service level agreements (SLA) [
3,
7], that is, the QoS in the service advertisement should be consistent with the QoS of the real service delivered to users. However, due to the variable network environments, many efforts should be devoted for service providers to improve the QoS and achieve the QoS level as they promised in the SLA. The QoS-aware service selection has been intensively discussed in many papers [
8,
9]. Recently, many research works on QoS-aware service selection in wireless mobile networks and in cloud computing have been proposed [
10‐
13]. The success of these mobile systems lies in their ability to provide users with cost-effective services that have the potential to run anywhere, anytime, and on any device without (or with little) user attention [
14]. As far as network-level QoS is concerned, much work has been carried out for both wired networks (i.e., the Internet) and wireless mobile ad hoc networks (MANETs) and mobile cellular networks. For instance, originated for the best-effort wired Internet, Integrated Service (IntServ) provides guaranteed bandwidth for each flow whereas Differentiated Services (DiffServ) offers guarantees on a per service class basis [
15]. In the current LTE 4G network, there are 9 bearer types that are used to achieve the different QoS levels for different client application services [
6]. While the device-to-device (D2D) communications as a more practical paradigm compared to conventional ad hoc networks, and a more cost-efficient paradigm compared to cellular networks, are advocated by many researchers [
16‐
18]. On the other hand, the cloud data centers can provide computing resources on users’ demands and the resources are almost infinite compared to the mobile devices. Thus, tenanting the service in cloud or transmitting the service with data to cloud to execute the computing in data centers is an efficient way to gain the powerful computing capability conveniently. The latter is usually called as offloading [
19]. With the prevailing of mobile communication techniques and devices, wireless accessing of cloud data centers, also called as mobile cloud computing, has become the next hotspot both in academia and industry. In the context of this paper, the
data service is specifically referred to a category of services which transmit users’ files or data through relay nodes to cloud data center for the purpose of cloud computing and cloud storage.
Not only the software as a service (SaaS) can gain the profit from cloud computing, the cloud platform can be beneficial to radio access networks (RANs) as well, which is an arising communication technology known as cloud radio accessing networks (C-RANs) [
3‐
5,
20‐
24]. Unlike conventional RANs, the C-RANs decouple the baseband processing unit (BBU) from the remote radio head (RRH), allowing for centralized operation of BBUs and scalable deployment of lightweight RRHs as small cells [
20]. BBUs locate in the signal processing cloud with high-speed fronthaul links to the distributed RRHs. The signal processing cloud is connected to backbone network by backhaul links. C-RAN is being advocated both by operators (e.g., China Mobile, SoftBank) as well as service providers (e.g., LightRadio, Liquid Radio). C-RAN is going to be the core technique in the next-generation broadband wireless networks.
In this paper, a QoS-aware data service providing framework is put forward. This framework is based on the C-RAN. The queuing theory is used to theoretically analyze the framework, and optimal service providing strategy is proposed for providers, which aims to minimize the cost for running a candidate data service which is subject to the execution duration constraint. At last, a simulation is conducted to do an empirical study for the proposed framework.