Search

Internet of Things (IoT) is envisioned to expand Internet connectivity of the physical world, and the mobile edge cloud can be leveraged to enhance the resource-constrained IoT devices. The performance of the cloud-enhanced IoT applications depends on various system-wide information, such as the wireless channel states between IoT devices and their corresponding serving edge cloud nodes. However, with the semi-trusted edge resources and the public nature of wireless channels, public sharing of system information should be avoided to better balance the tradeoff between performance and security. In this paper, the benefits of local information exchange is investigated, where the privately-owned physical layer channel information is leveraged to extract lightweight keys. For the point-to-point wireless communications links with multiple passive eavesdroppers, the security metric in terms of conditional min-entropy is evaluated via the proposed Dynamic Bayesian Model. The proposed model can flexibly incorporate various dynamic information flows in the system and quantify the information leakage caused by wireless broadcasting. The rigorously defined and derived security metrics for such a key generation pipeline has been verified via the real-world collected time-varying wireless channel data. The designed model can achieve previously inconceivable security properties.

Drone mounted base stations (DBSs) can be flexibly deployed over some areas (such as hotspots and disaster struck areas) to speed up the communication between terrestrial base stations (TBSs) and users. In order to increase the capacity of the backhaul link between a DBS and its TBS, free space optical (FSO) communications is applied as the backhaul solution. However, the DBS and its TBS should be in line-of-sight (LoS). Thus, a new DBS placement method is introduced to maximize the number of served users while guaranteeing the LoS between the DBS and its TBS. In addition, in order to prolong the hovering time of a DBS, we propose to deploy an optical beam to facilitate simultaneous communications and charging. That is, an optical beam transmitted from a TBS to its DBS carries not only data but also energy. The DBS mounted with a solar panel and an FSO receiver receives both the data and energy carried by the optical beam simultaneously. Numerical results demonstrate that using an optical beam to charge and communicate with a DBS can gain 25% extra hovering time of the DBS and achieve high network throughput.

Drone-mounted base-stations (DBSs) are promising complements and substitutes of the terrestrial base-stations. We investigate the problem of Placement and cOmmunications in the DBS-aided heterogeneous network (POD) with tier-1 backhaul using free space optics communications and tier-2 backhaul utilizing in-band full-duplex communications to provide ubiquitous connections and high spectrum efficiency. The POD problem is shown to be NP-hard, and thus we simplify the POD problem by decomposing it into three sub-problems. Finally, we propose an approximation algorithm to solve the POD problem, and demonstrate that the proposed algorithm is superior to two baseline algorithms. Keywords: Drone-mounted base-stations, Unmanned aerial vehicles, Heterogeneous networks, Wireless backhauling, Full-duplex, OFDMA, Resource allocation

Widely deployed smart cameras are generating a large amount of video data and capable of processing frames on devices. Empowered by edge computing, the video data can also be offloaded to edge servers for processing. By leveraging the on-device processing and computation offloading, we propose a federated video analytics system named FedVision to efficiently provision video analytics across devices and servers. The challenge of designing FedVision is to optimally use the computing and networking resources for video analytics. Since there is no closed-form expression of the system performance, black-box optimization is employed to optimize the system performance. However, using black-box optimization directly incurs excessive system queries that lead to very poor system performance. To solve this problem, we design a new optimization method that integrates black-box optimization with Neural Processes (NPs) as a system performance approximator. This method allows black-box optimizer to query NPs instead of the real system. We validate the performance of FedVision and the new optimization method using both numerical results and experiments with a testbed.

Drone-mounted base stations (DBSs) are promising means to provide ubiquitous connections to users and support various emerging applications in mobile networks while full duplex communications has the potential to improve the spectrum efficiency. In this paper, we investigate the backhaul-aware uplink communications in a full-duplex DBS-aided HetNet (BUD) problem with the objective to maximize the total throughput of the network while minimizing the number of deployed DBSs. Since the BUD problem is NP-hard, it is then decomposed into three sub-problems: the joint UE association, power and bandwidth assignment (Joint-UPB) problem, the DBS placement problem (including the vertical and horizontal positions) and the problem of determining the number of DBSs to be deployed. We propose two approximation algorithms to solve the first two sub-problems and use the linear programming to solve the last sub-problem. Finally, another approximation algorithm, referred to as the AA-BUD algorithms, is proposed to solve the BUD problem with guaranteed performance based on the solutions of the three sub-problems. The performance of the AA-BUD algorithm has been demonstrated via extensive simulations, and is shown to be superior to two benchmark algorithms with up to 62% throughput improvement.

Mobile edge computing (MEC) is emerging as a promising computing architecture by pushing computation and storage closer to end-users with both strategically deployed edge servers and opportunistic available resources. Essentially different from the traditional definition of cloud computing, mobile edge computing can provide solutions to problems faced by mobile cloud computing and support a large number of different mobile and the Internet of Things (IoT) applications and services. However, to maximize the resource utilization and user service experiences, it is envisioned that mobile edge computing and cloud computing should be designed and operated under a seamless integrated hierarchical architecture, though there are still many unprecedented technical challenges for applying mobile edge computing to the emerging mobile and IoT applications.

The fluctuation of received signal strength (RSS) induced by changing environment is the main hindrance from practical applications of the fingerprint-based indoor positioning methods. Transfer learning can mitigate the fluctuation of RSS by transferring knowledge from a source domain (off-line RSS data) to a target domain (online RSS data). However, the existing transfer learning approaches do not fully take into account the full constraints in Global and LOcal Structural conSistency (GLOSS), thus resulting in insufficient knowledge transfer. To overcome the above drawback, we propose a Transferred knowlEdge-Aided POsiTioning (TEAPOT) approach via GLOSS constraints in this paper. TEAPOT imposes the global structural consistency by minimizing the differences between the marginal and conditional distributions of the source and target domains and maximizing the samples variance in a latent subspace. Simultaneously, it also imposes the local structural consistency by minimizing within class variance and maximizing between class variance to retain the source discriminative information and preserving the local neighborhood relationship by using manifold regularization. Furthermore, a nonlinear TEAPOT is derived to improve the ability of TEAPOT to alleviate the limitation of linear projection. Compared with the existing methods, two of our proposed TEAPOT approaches via GLOSS constraints show higher accuracy and better ability in handling out-of-sample generalization. The experimental results verify that the proposed method significantly outperforms the existing methods.

Ram Adapa (Fellow, IEEE) received the B.S. degree from Jawaharlal Nehru Technological University, India, the M.S. degree from IIT Kanpur, India, and the Ph.D. degree from the University of Waterloo, ON, Canada, all in electrical engineering.

In this paper, we address the target device localization problem in the asynchronous networks. For the purpose of saving power resources, the target device is not synchronized with the anchor nodes, but is only required to listen to the signals transmitted from the anchors, which, however, introduces two extra nuisance parameters: the target's clock skew and clock offset. By transforming the time-of-arrival measurements into time-difference-of-arrival measurements, the clock offset of the target's clock is eradicated. However, there still exists the unknown clock skew, which may degrade the localization performance. Since the range of the clock skew is usually known as a priori, we assume that it follows a uniform distribution within this range. By doing so, we take it as a part of measurement noise and estimate the target node position only. To estimate the target node position, we formulate a fractional programming problem and further show that it can be solved by solving one single mixed semidefinite and second-order cone program (SD/SOCP). Simulation results illustrate the superior performance of the proposed method over the existing methods.

We propose a localization technique by fusing multiple classifiers based on received signal strengths (RSSs) of visible light in which different intensity-modulated sinusoidal signals emitted by LEDs are captured by photodiodes placed at various grid points. First, we obtain some approximate RSSs fingerprints by capturing the peaks of power spectral density of the received signals at each given grid point. Unlike the existing RSSs-based algorithms, several representative machine learning algorithms are adopted to train multiple classifiers based on these RSSs fingerprints. Then, two robust fusion localization algorithms, namely, grid-independent least square and grid-dependent least square (GD-LS), are proposed to combine the outputs of these classifiers. A singular value decomposition (SVD)-based LS (LS-SVD) method is proposed to mitigate the numerical stability problem when the prediction matrix is singular. Experiments conducted on an intensity-modulated direct detection system show that the probability of having mean square positioning error of less than 5 cm achieved by GD-LS is improved by 93.03% and 93.15%, respectively, as compared to those by the RSS ratio and RSS matching methods with the fast Fourier transform length of 2000.

In this paper, we consider the problem of resource allocation in a dense small-cell network. Each small-cell base station is powered by a renewable energy source and operates in the full-duplex mode. We account for the rate-dependent energy term for data decoding into the total energy consumption at the small-cell base station. Owing to this new energy term, the transmitter and receiver operations now draw the energy from a common source. For a new energy consumption model and high interference scenario, which arises due to full-duplex communications, we formulate an energy and load aware resource management optimization problem under the energy causality and total transmit power constraints of the small-cell base station and uplink user equipments. In particular, the problem minimizes the data queue length of each network user equipment by jointly designing the beamformers, power, and sub-carrier allocation and their scheduling. Owing to the non-convexity of the problem, a global solution is inefficient; thus, we opt for the successive parametric convex approximation method to obtain a sub-optimal solution. This method solves for the convex approximate of the non-convex problem in each iteration and leads to faster convergence. For practical implementation, we further develop a distributed algorithm by using the dual decomposition framework, which relies on limited exchange of information between the involved base stations. Numerical simulations compare the network scenario which accounts for uplink channel rate-dependent energy consumption with that which ignores it. Results advocate the need for redesigning of the resource allocation scheme. In addition, numerical simulations also validate the usefulness of full-duplex communications over the half-duplex communications in terms of minimizing the sum data queue length of the users.

Matrix inversion is a fundamental operation for solving linear equations for many computational applications, especially for various emerging big data applications. However, it is a challenging task to invert large-scale matrices of extremely high order (several thousands or millions), which are common in most Web-scale systems, such as social networks and recommendation systems. In this paper, we present an lower upper decomposition-based block-recursive algorithm for large-scale matrix inversion. We present its well-designed implementation with optimized data structure, reduction of space complexity, and effective matrix multiplication on the Spark parallel computing platform. The experimental evaluation results show that the proposed algorithm is efficient to invert large-scale matrices on a cluster composed of commodity servers and is scalable for inverting even larger matrices. The proposed algorithm and implementation will become a solid foundation for building a high-performance linear algebra library on Spark for big data processing and applications.

An important issue in the cellular industry is the rising energy cost and carbon footprint due to the rapid expansion of the cellular infrastructure. Greening cellular networks has thus attracted attention. Among the promising green cellular network techniques, the renewable energy-powered cellular network has drawn increasing attention as a critical element towards reducing carbon emissions due to massive energy consumption in the base stations deployed in cellular networks. Game theory is a branch of mathematics that is used to evaluate and optimize systems with multiple players with conflicting objectives and has been successfully used to solve various problems in cellular networks. In this paper, we model the green energy utilization and power consumption optimization problem of a green cellular network as a pilot power selection strategic game and propose a novel distributed algorithm based on a strategic learning method. The simulation results indicate that the proposed algorithm achieves correlated equilibrium of the pilot power selection game, resulting in optimum green energy utilization and power consumption reduction.