Flexible resource allocation adaptive to communication strategy selection for cellular clients using Stackelberg game

Abstract

In cooperative communication Networks (or cooperative cognitive radio networks), a source node usually recruits non-altruistic relays to execute cooperative communication with a target node, through compensating these potential relays with a fraction of its licensed frequency band. When a source node is far apart from a target node, it is difficult to achieve the desired transmission performance by cooperative communication. In this paper, a source node's communication strategy can dynamically be adjusted with help of a base station according to variation of wireless communication environments. Also, an appropriate relaying path can be constructed to replace a long-distance wireless link without demand of channel state information. Furthermore, following the proposed Stackelberg game-theoretic method based on three-party-cooperation, a base station can determine the preliminary division of a source node's licensed frequency band and the initial power allocation for all the relays in the same relaying path, while the source node and the related relays can dynamically adjust them respectively during data transmission according to information feedback.

Introduction

With the aid of multi-hop relays, conventional cellular networks can improve coverage area and increase the capacity [1], [2]. Furthermore, the low power base stations are introduced to cellular systems to enhance coverage and improve system capacity, which is considered in the standardization process of the next generation cellular network, such as Heterogeneous Cellular Networks (HCNs) [3], [4], [5]. Meanwhile, Multi-hop Heterogeneous Cellular Networks (MHCNs) [6] are also the tendency of future wireless networks.

However, network operators must make a huge investment in the installation and maintenance of the low power base stations. The relays deployed by network operators require flexible site acquisition due to the low transmission range, which is important but intractable for network service providers.

In some outdoor scenarios (e.g., sports meetings, fairs, concerts, and festivals), it is an economical and effective approach to only select wireless clients to serve as relays, which can both improve coverage area and increase the capacity. A wireless client device at a cell edge (or poor coverage area) can communicate with the base station through relaying its information via other devices.

However, the increased benefit comes at the expense of consumption of the limited resources of the wireless clients serving as relays. Since the potential relays may belong to different network entities (or operators), they are usually selfish nodes, and thus unwilling to cooperate without any additional incentive.

It is widely employed in Cooperative Communication Networks (CCNs) and Cooperative Cognitive Radio Networks (CCRNs) to regard a part of one node's licensed frequency band as a reward to stimulate the cooperative relaying behaviors of other nodes. However, the related solutions are not suitable for the above mentioned application scenario. For example, for a wireless client device located in a macro-cell edge, it cannot obtain a desired link quality to communicate with the base station by the cooperative communication mode, since the quality of link between the sender and the relays (or that between the relays and the base station) may be bad due to the long transmission distance and the possible blockage of buildings (or obstacles).

Usually, multiple devices play the role of relays for a device at a cell edge, which can reduce the transmission distance of links in the path from this device to the base station. Unlike the multiple relays in the CCNs or CCRNs that concurrently relay the data from the source to the destination (i.e., the cooperative relaying mode or multiple relaying mode, as discussed in [7]), we focus on the case that the data from a source is relayed to a destination by several relays one by one (i.e., the multi-hop relaying mode), where the relays are all located in the path with at least three hops from a source to a destination. In essence, multi-hop relaying is a serial relaying mode while cooperative or multiple relaying is a parallel relaying mode. Furthermore, in a serial relaying path, each link may employ one of direction communication, cooperative relaying or multiple relaying modes according to link quality status.

In CCNs, relaying nodes can act as a virtual antenna array to help the source node forward its information to the destination node, and thus the performance depends on careful resource allocations (e.g., relay placement, relay selection, and power control). The authors in [8] provide the analysis on symbol error rates and optimum power allocations for the decode-and-forward cooperation protocol in wireless networks, while the authors in [9] focus on when to cooperate and which relay to cooperate with, which requires Channel State Information (CSI) during relay selection. Wang et al. [10] use Stackelberg game to design relay selection and power allocation schemes in CCNs. In [11], the authors propose a two-hop spectrum leasing scheme, which allows the source to communicate through using both cooperative communication and direct transmission in the same slot of CCNs.

In CCRNs, if a Secondary User (SU) is oblivious to a Primary User (PU), the acquisition of unused licensed spectrum bands is not guaranteed. Spectrum leasing has been proposed to tackle this problem through negotiation between the PU and SU, which allows the SU to acquire unused licensed spectrum band for a guaranteed period of time.

There are the various conditions for spectrum leasing. Some examples are as follows. In [12], a PU leases a part of its time slots to the SUs if its data rate in cooperative communication is higher than that in direct transmission, while a PU in [13], [14] determines to do so if the payment is to satisfy the quotation. For the work in [15], a PU leases a part of the bandwidth aiming at reducing its power consumption through the cooperation of SUs, but the certain constraint on the Bit Error Rate (BER) of both a PU and a SU is requested. The work in [16] is to maximize both the power saving of a PU and the transmission rate of a SU on the leased bandwidth, while a PU's main purpose in [17] is to maximize its power saving by exploiting the cooperation from a SU under the condition that its transmission rate is guaranteed. The authors in [18] consider the power grid to design their Stackelberg game scheme, so cognitive networks sense not only the radio spectrum environment but also the smart grid environment to determine a quotation.

Unlike the aforementioned works based on the Stackelberg game, which only consider the strategy space with a single action, the schemes in [19], [20] are modeled as the sum-constrained power allocation game with two bounded powers coupled together [21]. However, they usually have the relatively high selection criteria for relays, for example, the relays have to use the granted frequency bands timely, and perceive the heterogeneity of the channel gains of users. The former may make the potential relaying nodes very depressed, while the latter will exclude a large number of ordinary nodes to act as relays.

Different from the existing spectrum leasing works based on Stackelberg game, our work addresses the problem formulation in the multi-hop relaying mode with respect to spectrum leasing and power allocation by Stackelberg game. The key results and contributions are included as follows.

• Our scheme can construct an appropriate relaying path for the multi-hop relaying mode, where the overhead of path construction can be reasonably shared among the base station, the source node, and the related relays.

• Since our scheme does not depend on CSI, it is more adaptive to rapid variations of wireless channels. Moreover, if the relaying nodes cannot timely use the granted frequency bands, they will get some compensation according to a setting strategy.

• Based on Stackelberg game model [22], we devise a three-party-cooperation method for resource allocation, where the base station can determine the preliminary division of the licensed frequency band of the source node and the power allocation of all the relays in the same relaying path, while the source node and the related relays can dynamically adjust them respectively during data transmission according to the information feedback.

The rest of the paper is organized as follows. In Section 2, we describe the system model and problem statement. Our scheme is presented in detail in Section 3. An analytic evaluation of our scheme and the most relevant work is given in Section 4, while the simulation results and analysis are presented in Section 5. Finally, Section 6 draws the conclusions.