ABSTRACT Recently, Multi‐label learning has emerged as a crucial technique across various fields. However, it typically requires a large amount of labelled data for effective model training. To address this limitation, many researchers have introduced the label propagation (LP) algorithm from semi‐supervised learning into multi‐label learning. Unfortunately, conventional LP approaches only transform multi‐label problem into single‐label formulations, thereby ignoring label correlations. This paper proposes a new LP algorithm integrating fuzzy C‐means (FCM) and sorting empowerment for multi‐label learning, termed C lustering‐based S orting Empowerment L abel P ropagation (CSLP). Firstly, FCM is employed to strengthen the relationship among labels, ensuring that the probability results of the LP algorithm in multi‐label scenarios are more consistent with actual data distributions. Afterward, sorting empowerment, inspired by group decision theory, is adopted as an ensemble strategy to enhance the robustness of the LP algorithm. Finally, extensive experiments on seven multi‐label datasets demonstrate that CSLP outperforms competing algorithms across four evaluation metrics (Hamming loss, Average precision, Recall score, and MicroF1).

With the development of intelligent technology, data exhibits characteristics of multi-label and imbalanced distribution, which lead to the degradation of classification model performance. Therefore, addressing multi-label class imbalance has become a hot research topic. Nowadays, over-sampling approaches aim to generate a superset of the original dataset to deal with imbalanced data. However, traditional over-sampling methods only employ the central data point and its nearest neighbor samples to synthesize samples without considering the impact of data distribution. To address these issues, in this paper, we propose an ensemble multi-label over-sampling algorithm (MLCIO) based on Chebyshev inequality and a group optimization strategy. Firstly, to generate more representative and diverse samples, with the seed sample serving as the sphere’s center, Chebyshev inequality is utilized to ensure that synthetic samples fall within its m times the standard deviation. Secondly, a group optimization ranking weighting approach is employed to obtain more reliable and stable label information. Finally, comparative experiments are conducted on 11 imbalanced datasets from various domains using different evaluation metrics. The results demonstrate that our proposal achieves better performance than other approaches.

The resolution of proof-of-work problem in blockchain requires significant amount of resources, while the lack of computing power on mobile devices limits the development of blockchain in mobile applications. To mitigate this issue, the combination of blockchain and mobile edge computing (MEC) has attracted much attention. In this paper, we consider an edge-enabled blockchain system that includes one single edge service provider (ESP), multiple types of miners and edge nodes. Each miner submits offloading request to ESP. In response, ESP designs contract to incentivize various types of edge nodes to contribute resources and offer computational services to the miners. This problem is a joint optimization problem of offloading decisions and contract design. Due to the time-variability of network environment, the randomness of miners’ task demands, and the asymmetric information between the ESP and edge nodes, solving this problem is challenging. We propose a deep reinforcement learning contract mechanism (DRLCM) for incentive-based computation offloading strategies, which divides the original problem into two sub-problems: computation offloading and contract design. Initially, the deep Q-network (DQN) algorithm is used to update the offloading decisions based on the evolving task demands and network conditions. Secondly, contract is designed to motivate edge nodes to participate in resource sharing. The problem is simplified by analyzing the necessary and sufficient conditions of feasible contract, and the Lagrange multiplier method is used to approximate the optimal contract. Simulation experiments demonstrate the effectiveness of the DRLCM algorithm, which shows better convergence and performance compared to traditional DQN, Double-DQN algorithms and Dueling-DQN.

In this paper, we aim to improve the throughput capacity of a wireless powered network by allowing user devices (UDs) to adapt channel access strategies. A base station (BS) can receive data and provide RF energy for UDs simultaneously in a full duplex mode. Each UD can choose a flexible access mode to transmit its data with the non-orthogonal multiple access technique or assist the other UDs' data transmissions via backscattering when it has less urgent data demands or insufficient energy supply. We maximize the sum throughput by optimizing the UDs' channel access mode, time allocation and beamforming strategies, according to the UDs' channel conditions and energy status. Practically, maximizing throughput is a challenging problem due to the uncertain channel information, the UDs' dynamic traffic demands and limited energy storages. Therefore, we propose a hierarchical learning approach to decompose the access mode selection and transmission control in two steps. We first employ a multi-agent deep reinforcement learning approach to update each UD's channel access strategy by interacting with the uncertain network environment. Then, we efficiently update the UDs' time allocation and the BS's beamforming strategies to further enhance the sum throughput. The simulation results verify a significant improvement in terms of the throughput and the learning efficiency compared to the benchmark methods.

Mobile edge computing (MEC) is viewed as a promising technology to address the challenges of intensive computing demands in hotspots (HSs). In this paper, we consider a unmanned aerial vehicle (UAV)-assisted backscattering MEC system. The UAVs can fly from parking aprons to HSs, providing energy to HSs via RF beamforming and collecting data from wireless users in HSs through backscattering. We aim to maximize the long-term utility of all HSs, subject to the stability of the HSs' energy queues. This problem is a joint optimization of the data offloading decision and contract design that should be adaptive to the users' random task demands and the time-varying wireless channel conditions. A deep reinforcement learning based contract incentive (DRLCI) strategy is proposed to solve this problem in two steps. Firstly, we use deep Q-network (DQN) algorithm to update the HSs' offloading decisions according to the changing network environment. Secondly, to motivate the UAVs to participate in resource sharing, a contract specific to each type of UAVs has been designed, utilizing Lagrangian multiplier method to approach the optimal contract. Simulation results show the feasibility and efficiency of the proposed strategy, demonstrating a better performance than the natural DQN and Double-DQN algorithms.

The resource allocation problem in Multi-access Edge Computing (MEC) has been widely studied to maximize its operation efficiency under limited resource constraint. However, the existing literatures overlooked the setup cost and the associated dynamic operations. In this work, we consider server configuration and overload in the multi-server scenario where servers are switched on/off depending on the network environment. A novel pricing mechanism maximizing the utility of base station (BS) monitoring multiple servers is proposed, which jointly optimizes the setup cost and server load. We aim to maximize the BS utility under one-day task requests, and divide the time into off-peak and peak periods based on task requests. In the off-peak period, we flexibly switch on/off servers for BS to reduce setup costs. In the peak period, to avoid overloading, we introduce crowdsourcing where servers as agents purchase idle resources from private users (PUs) for mobile users (MUs) and minimize MUs’ cost by a contract-based knapsack algorithm. Lastly, a pricing mechanism is proposed to solve the BS utility maximization problem with an exploratory Upper Confidence Bound (UCB)-based algorithm adjusting server prices dynamically. Simulation results show that the proposed algorithm is superior to others in minimizing MUs cost and maximizing BS utility.

In this paper, we consider a mobile edge computing (MEC) system with multiple unmanned aerial vehicles (UAVs) and stochastic energy harvesting. The UAVs' mobility can help data offloading over a larger geographical area containing multi- hotspots (HSs). If HSs have offloading requests, the dispatch agent (DA) can recruit different types of UAVs to fly close to HSs and help computation. We aim to maximize the long-term utility of all HSs, subject to the stability of energy queue. The proposed problem is a joint optimization problem of offloading strategy and contract design in a dynamic setting over time. We design a deep reinforcement learning based contract incentive (DRLCI) strategy that solves the joint optimization problem in two steps. Firstly, we use an improved deep Q-network (DQN) algorithm to obtain the offloading decision. Secondly, to motivate UAVs to participate in resources sharing, a contract has been designed for asymmetric information scenarios, and Lagrangian multiplier method has been utilized to approach the optimal contract. Simulation results show the feasibility and efficiency of the proposed strategy. It can achieve a very close-to the performance obtained by complete information scenario.

In this paper, we employ multiple unmanned aerial vehicles (UAVs) to assist sensing data transmission from the ground users (GUs) to the remote base station (BS). Each UAV can first cache the sensing data and then report the cached data to the BS. We consider a time-slotted protocol to coordinate the UAVs' data collection and reporting. Only one UAV is allowed to forward its data to the BS in each time slot. We formulate a multi-stage stochastic optimization problem to minimize the longterm age-of-information (AoI) by jointly optimizing the UAVs' trajectories and scheduling strategies. To simplify this problem, we model the dynamics of the UAVs' data buffer and AoI statuses by queueing systems, and propose a novel AoI-aware Adaptation scheme. This scheme allows us to transform the multistage dynamic programming problem into per-slot scheduling and trajectory planning sub-problems by using the Lyapunov optimization framework. Then, in each time slot, we can update the UAVs' scheduling and flying strategies in an iterative manner according to the instant buffer and AoI statuses. Simulation results show that the proposed scheme outperforms baseline schemes in terms of reducing AoI while stabilizing and balancing the UAVs' data queues.

In this paper, we consider a hybrid computation offloading scheme that allows edge users to offload workloads to the edge servers by using active RF communications and backscatter communications. We aim to maximize the overall energy efficiency by jointly optimizing the beamforming of access point (AP) and the users’ offloading decisions. Considering a dynamic environment, we propose a hierarchical multi-agent deep reinforcement learning (H-MADRL) framework to solve this problem. The high-level agent resides in the AP and optimizes the beamforming strategy, while the low-level user agents learn and adapt individuals’ offloading strategies. To further improve the learning efficiency, we propose a novel optimization-driven learning algorithm that allows the AP to estimate the low-level users’ actions by solving an approximate problem efficiently. Then, the action estimation can be shared with all users and drive them to update individuals’ actions independently. Simulation results reveal that our algorithm can improve the reward performance by 50%. The learning efficiency and reliability are also enhanced comparing to the conventional model-free learning methods.

Here, the authors study the novel problem of optimising access point (AP) association by maximising the network throughput, subject to the degree bound of AP. The formulated problem is a combinatorial optimisation. They resort to the Markov Chain approximation technique to design a distributed algorithm. They first approximate their optimal objective via Log‐Sum‐Exp function. Thereafter, they construct a special class of Markov Chain with steady‐state distribution specify to their problem to yield a distributed solution. Furthermore, they extend the static problem setting to a dynamic environment where the users can randomly leave or join the system. Their proposed algorithm has provable performance, achieving an approximation gap of . It is simple and can be implemented in a distributed manner. Their extensive simulation results show that the proposed algorithm can converge very fast, and achieve a close‐to‐optimal performance with a guaranteed loss bound.

The heterogeneous network structure is a promising paradigm to improve the quality of service across the entire network. Nevertheless, such a structure is challenging due to the presence of multiple‐tier secondary users (SUs). In this study, the authors investigated the effect of spectrum allocation in heterogeneous cognitive radio networks with a primary network and two‐tier secondary networks, and proposed a two‐tier spectrum trading strategy which includes two trading processes. In Process One, they model the spectrum trading as a monopoly market, where the primary spectrum owner (PO) acts as the monopolist and the first‐tier secondary users (FSUs) act as the buyers. They design an optimal quality‐price contract to maximise the utility of PO, and the FSUs will choose the spectrum with appropriate quality and price to enhance their satisfaction. In Process Two, spectrum trading is modelled as a multi‐seller, multi‐buyer market. The dynamic behaviour of second‐tier SUs is studied using the theory of evolution game, while the competition among FSUs is analysed via a non‐cooperative game where the Nash equilibrium is considered as the solution. The existences of the optimal contract, evolutionary equilibrium and Nash equilibrium are demonstrated in the performance evaluation.

The indoor positioning method based on received signal strength indication (RSSI) ranging is a lack of systematic quantitative research on the factors affecting the characteristics of indoor RSSI, which affect the accuracy of positioning. This paper quantitatively analyzes the characteristics of the 2.4 GHz RSSI collected in indoor scenes from the following four aspects: antenna orientation of the receiver, type of wireless network interface card (NIC) of the receiver, time period of the data measurement and height difference between the transmitting and receiving antenna. The experimental results show that: (i) the RSSI value measured is the strongest when the antenna of the receiver is vertically oriented to the antenna of the transmitter, while the weakest when the antenna is vertically back‐facing and the difference between the strongest signal and the weakest signal is 20–25% at the same test point; (ii) the wider the measurement range of NIC, the more conducive to data collection; (iii) the distribution of RSSI signals generated by an access point at a fixed position is inconsistent with time; (iv) when the height of receiving antenna is 0.5 m different from each other, the path loss index of ranging model produces a deviation of about 10%.

Classification is one of the most important problems in data mining and machine learning. The quality and quantity of classification rules are two factors to influence the accuracy of classification. In this paper, we propose a new algorithm to enhance the final classification accuracy, called CMCM (Classification based on Multiple Classifier Models), which consists of two classification models. Model1 centers on the improvement of quality. The optimal attribute values are obtained as the first item of a classification rule from both the items and their complements. While in Model2, quantity is taken into consideration, so it constructs two candidate sets and uses the one-versus-many strategy to generate several rules at one time. The experiment results demonstrate that CMCM can achieve higher classification accuracy than the proposed classification approaches. CMCM can extract sufficient high-quality rules for imbalanced data. Meanwhile, it can also obtain sufficient latent information for classification.

In 2008, the Federal Communications Commission issued a ruling permitting the unlicensed usage of TV white spaces (TVWS), i.e. locally vacant TV channels. Due to its low‐frequency range (50–698 MHz), the TV spectrum has much better propagation characteristic and higher‐spectral efficiency, resulting in a wide range of potentially important applications. However, unlike typical cellular and industrial scientific medical bands, TVWS are subjected to high‐spatial variation, temporal variation, and fragmentation, resulting in new challenges in TVWS identification and in implementing a wireless network in this band. Identification and network design are the two key issues required to be addressed while investigating TVWS. These two problems have been widely discussed in several existing literature. Applications in TVWS are also an important topic, which has not been adequately explored. This study provides an up‐to‐date survey of TVWS and its applications in future wireless networks and communication. Various problems and challenges associated with each use case as well as the possible enabling methods to address these challenges are also presented.

One important issue associated with spectrum management in heterogeneous cognitive radio network is: how to appropriately allocate the spectrum to the secondary senderdestination (S-D) pair for sensing and utilization. In this work, the authors investigate the spectrum allocation problem under a more practical scenario, taking the heterogeneous characteristics of both secondary S-D and PU channels into consideration. With the objective to maximize the achievable throughput for secondary S-D, we formulate the spectrum allocation problem as a linear integer optimization problem under spectrum availability constraint, spectrum span constraint and interference free constraint. This problem is NP-complete, and a recent result in theoretical computer science called randomized rounding algorithm with polynomial computational complexity is developed to find the ρ-approximation solution. Evaluation results show that our proposed algorithm can achieve a close-to-optimal solution while keeping the complexity low.

Spectrum sensing and access have been widely investigated in cognitive radio network for the secondary users to efficiently utilize and share the spectrum licensed by the primary user.We propose a cluster-based adaptive multispectrum sensing and access strategy, in which the secondary users seeking to access the channel can select a set of channels to sense and access with adaptive sensing time.Specifically, the spectrum sensing and access problem is formulated into an optimization problem, which maximizes the utility of the secondary users and ensures sufficient protection of the primary users and the transmitting secondary users from unacceptable interference.Moreover, we explicitly calculate the expected number of channels that are detected to be idle, or being occupied by the primary users, or being occupied by the transmitting secondary users.Spectrum sharing with the primary and transmitting secondary users is accomplished by adapting the transmission power to keep the interference to an acceptable level.Simulation results demonstrate the effectiveness of our proposed sensing and access strategy as well as its advantage over conventional sensing and access methods in terms of improving the achieved throughput and keeping the sensing overhead low.

Cognitive radio is proposed as an efficient way to address the issue of spectrum shortage and under- utilization, in which cooperative spectrum sensing (CSS) is used to enhance the sensing performance. One of the most fundamental problems of CSS is: how to appropriately assign the secondary users (SUs) to sense the primary user (PU) channels? In this paper, We study the CSS problem under a more practical scenario where taking the heterogeneous characteristics of both SUs and PU channels into consideration. With the objective to maximize the achievable throughput for SUs, we propose a cluster- based CSS to obtain a proper assignment policy, in which all the cluster members cooperative in sensing the same channels, moreover, the CSS problem is formulated as a Maximum Weight One-Sided Biclique Problem, and a greedy heuristic algorithm is proposed to find the suboptimal assignment policy. To evaluate the tradeoff between sensing accuracy and spectrum opportunity, the simulation is conducted between the number of sensed channels and the achievable throughput.

MAC protocols to sense channels for data transmission have been widely investigated for the secondary users to efficiently utilize and share the spectrum licensed by the primary user. One important issue associated with MAC protocols design is how the secondary users determine when and which channel they should sense and access without causing harmful interference to the primary user. In this paper, we jointly consider the MAC-layer spectrum sensing and channel access. Normal Spectrum Sensing (NSS) is required to be carried out at the beginning of each frame to determine whether the channel is idle. On detecting the available transmission opportunity, the secondary users employ CSMA for channel contention. The novelty is that, Fast Spectrum Sensing (FSS) is inserted after channel contention to promptly detect the return of the primary users. This is unlike most other MAC protocols which do not incorporate FSS. Having FSS, the primary user can benefit from more protection. A concrete protocol design is provided in this paper, and the throughput-collision tradeoff and utility-collision tradeoff problems are formulated to evaluate its performance. Simulation results demonstrate the efficiency of the proposed MAC protocol with FSS.

Mobile infostation networks which achieve content distribution by exploiting the opportunistic contact among mobile users (MUs) and access points (APs) have attracted a lot of attention in the last few years. It is found that cooperation either between APs or MUs has a great impact on the network performance. However, in most of the existing works wherever cooperation is involved, cooperation among the APs or MUs is discussed independently, while few of them has jointly taken both levels of cooperation into account. In this paper, we propose a more general framework which allows two levels of cooperation. The benefit as well as the cost of such hierarchical cooperation are taken into consideration. With properly defined payoffs of both APs and MUs, each AP may gain more benefit by strategically forming coalition with other APs, while the MUs can also choose to cooperate with one another to further maximize their payoffs. To obtain the optimal structure of two-level cooperation, an implementable distributed algorithm is proposed. Through extensive numerical experiments, our scheme shows the high effectiveness in achieving the stable cooperation structure, and the impact of different system parameters is also extensively investigated.

In this paper, optimal multiple channels cooperative spectrum sensing setting in non-identical environment is investigated. In previous work on cooperative sensing, all the secondary users have the same detection threshold and the noise received is independent identically distributed random variable. Thus researchers often assume that identical sensing time is assigned to the channels for spectrum sensing. In our paper, secondary users cooperatively sense the channel and send the binary results to the common receiver where energy detection with hard decision rule is employed. We assume that secondary users can assign individual sensing time to the channel with possibly different noise powers and detection thresholds. An iterative algorithm with polynomial complexity is established to determine the optimal sensing sequence for the secondary users to assign the mini-slots to the channels, such that the throughput increase can be maximized in each iteration. Furthermore, a new performance metric delay sensitivity is introduced to evaluate how long the authorized transmitting users need to wait for data transmission. Simulation results show that this iterative algorithm can yield better performance in cooperative spectrum sensing than other schemes with identical sensing setting in terms of both achievable throughput and delay sensitivity.

In this paper, we study the throughput and delay scaling laws of two coexisting mobile networks. By considering that both the primary and secondary networks are mobile and move according to random walk model, we propose a multi-hop transmission scheme. Based on the assumption that the secondary node can help to relay the primary packet, we show that the secondary network can achieve the same throughput and delay scaling laws as in stand-alone network Ds(m)=Θ(mλs(m)). Furthermore, for primary network, it is shown that the throughput-delay tradeoff scaling law is given by Dp(n)=Θ(√(n log n λp(n))), when the primary node is chosen as relay node. If the relay node is a secondary node, the scaling law is Dp(n)=Θ(√(nβlog n λp(n))), where β>;1. The novelties of this paper lie in: i) Detailed study of the delay scaling law of the secondary network in the complex scenario where both the primary and secondary networks are mobile; ii) The impact of buffer delay on the primary and secondary networks due to the presence of preservation region. We explicitly analyze the buffer delay and obtain an expression as DII(SR)(m)=Θ(1/√(nβ-1as(m))).