Privacy-Preserving Federated Mobility Prediction with Compound Data and Model Perturbation Mechanism

Human mobility prediction is important for many applications.However,training an accurate mobility prediction model requires a large scale of human trajectories,where privacy issues become an important problem.The rising federated learning provides us with a promising solution to this problem,which enables mobile devices to collaboratively learn a shared prediction model while keeping all the training data on the device,decoupling the ability to do machine learning from the need to store the data in the cloud.However,existing federated learningbased methods either do not provide privacy guarantees or have vulnerability in terms of privacy leakage.In this paper,we combine the techniques of data perturbation and model perturbation mechanisms and propose a privacy-preserving mobility prediction algorithm,where we add noise to the transmitted model and the raw data collaboratively to protect user privacy and keep the mobility prediction performance.Extensive experimental results show that our proposed method significantly outperforms the existing stateof-the-art mobility prediction method in terms of defensive performance against practical attacks while having comparable mobility prediction performance,demonstrating its effectiveness.

Routing Protocol for Heterogeneous FANETs with Mobility Prediction

In recent years,with the growth in Unmanned Aerial Vehicles(UAVs),UAV-based systems have become popular in both military and civil applications.In these scenarios,the lack of reliable communication infrastructure has motivated UAVs to establish a network as flying nodes,also known as Flying Ad Hoc Networks(FANETs).However,in FANETs,the high mobility degree of flying and terrestrial users may be responsible for constant changes in the network topology,making end-to-end connections in FANETs challenging.Mobility estimation and prediction of UAVs can address the challenge mentioned above since it can provide better routing planning and improve overall FANET performance in terms of continuous service availability.We thus develop a Software Defined Network(SDN)-based heterogeneous architecture for reliable communication in FANETs.In this architecture,we apply an Extended Kalman Filter(EKF)for accurate mobility estimation and prediction of UAVs.In particular,we formulate the routing problem in SDN-based Heterogeneous FANETs as a graph decision problem.As the problem is NP-hard,we further propose a Directional Particle Swarming Optimization(DPSO)approach to solve it.The extensive simulation results demonstrate that the proposed DPSO routing can exhibit superior performance in improving the goodput,packet delivery ratio,and delay.

New multicast scheme based on Dynamic mobility prediction in mobile IPv6 environment

A new mobile multicast scheme called mobility prediction based mobile multicast(MPBMM) was proposed. In MPBMM, when a mobile node (MN) roams among subnets during a multicast session, MN predicts the next subnet, to which MN will attach, by the information of its position and mobility speed, consequently speeds up the handoff procedure. Simulation results show that the proposed scheme can minimize the loss of multicast packets, reduce the delay of subnet handoff, decrease the frequency of multicast tree reconfiguration, and optimize the delivery path of multicast packets. When MN moves among subnets at different speeds (from 5 to 25 ms), the maximum loss ratio of multicast packets is less than0.2%, the maximum inter-arrival time of multicast packets is 117 ms, so the proposed scheme can meet the QoS requirements of real-time services. In addition, MPBMM can support the mobility of multicast source.

Interference management via access control and mobility prediction in two-tier heterogeneous networks

Abstract： Two-tier heterogeneous networks （HetNets）, where the current cellular networks, i.e., macrocells, are overlapped with a large number of randomly distributed femtocells, can potentially bring significant benefits to spectral utilization and system capacity. The interference management and access control for open and closed femtocells in two-tier HetNets were focused. The contributions consist of two parts. Firstly, in order to reduce the uplink interference caused by MUEs （macrocell user equipments） at closed femtocells, an incentive mechanism to implement interference mitigation was proposed. It encourages femtoeells that work with closed-subscriber-group （CSG） to allow the interfering MUEs access in but only via uplink, which can reduce the interference significantly and also benefit the marco-tier. The interference issue was then studied in open-subscriber-group （OSG） femtocells from the perspective of handover and mobility prediction. Inbound handover provides an alternative solution for open femtocells when interference turns up, while this accompanies with PCI （physical cell identity） confusion during inbound handover. To reduce the PCI confusion, a dynamic PCI allocation scheme was proposed, by which the high handin femtocells have the dedicated PCI while the others share the reuse PCIs. A Markov chain based mobility prediction algorithm was designed to decide whether the femtoeell status is with high handover requests. Numerical analysis reveals that the UL interference is managed well for the CSG femtocell and the PCI confusion issue is mitigated greatly in OSG femtocell compared to the conventional approaches.

Research on road topology based mobility prediction schemes

Geographic routing has been introduced in mobile ad hoc networks and sensor networks. But its per-formance suffers greatly from mobility-induced location errors that can cause Lost Link （LLNK） and LOOP problems. Thus various mobility prediction algorithms have been proposed to mitigate the errors, but sometimes their prediction errors are substantial. A novel mobility prediction technique that incorpo-rates both mobile positioning information and road topology knowledge was presented. Furthermore, the performance of the scheme was evaluated via simulations, along with two other schemes, namely, Linear Velocity Prediction （LVP） and Weighted Velocity Prediction （WVP） for comparison purpose. The results of simulation under Manhattan mobility model show that the proposed scheme could track the movement of a node well and hence provide noticeable improvement over LVP and MVP.

Participants Recruitment for Coverage Maximization by Mobility Predicting in Mobile Crowd Sensing

Mobile Crowd Sensing(MCS)is an emerging paradigm that leverages sensor-equipped smart devices to collect data.The introduction of MCS also poses some challenges such as providing highquality data for upper layer MCS applications,which requires adequate participants.However,recruiting enough participants to provide the sensing data for free is hard for the MCS platform under a limited budget,which may lead to a low coverage ratio of sensing area.This paper proposes a novel method to choose participants uniformly distributed in a specific sensing area based on the mobility patterns of mobile users.The method consists of two steps:(1)A second-order Markov chain is used to predict the next positions of users,and select users whose next places are in the target sensing area to form a candidate pool.(2)The Average Entropy(DAE)is proposed to measure the distribution of participants.The participant maximizing the DAE value of a specific sensing area with different granular sub-areas is chosen to maximize the coverage ratio of the sensing area.Experimental results show that the proposed method can maximize the coverage ratio of a sensing area under different partition granularities.

Smart Prediction for Seamless Mobility in F-HMIPv6 Based on Location Based Services

This paper proposed a new scheme for smooth handoff over F-HMIPv6 networks based on Location Based Services(LBS). It uses the available information about mobile node(MN) such as user mobility patterns and MN's velocity to reduce handoff latency. In the proposed scheme, the movement pattern of users plays an important role in the performance analysis. The proposed scheme develops LBS that records a number of mobility patterns based on user daily behavior, and exploits these information to reduce the handoff latency. The proposed scheme is analyzed and compared with F-HMIPv6 protocol. The results show that it improves the performance in terms of handoff latency, packet delivery cost, and location update cost. Specifically, the proposed scheme achieves the tradeoff between realizing a smooth handoff and minimizing the cost that is the demand for all customers of mobile services.

Theoretical Prediction of the Photovoltaic Properties of BFBPD-PC61 BM System as a Promising Organic Solar Cell

In this work,the photovoltaic properties of BFBPD-PC61 BM system as a promising high-performance organic solar cell（OSC） were theoretically investigated by means of quantum chemistry and molecular dynamics calculations coupled with the incoherent charge-hopping model.Moreover,the hole carrier mobility of BFBPD thin-film was also estimated with the aid of an amorphous cell including 100 BFBPD molecules.Results revealed that the BFBPD-PC61 BM system possesses a middle-sized open-circuit voltage of 0.70 V,large short-circuit current density of 17.26 mA ·cm^-2,high fill factor of 0.846,and power conversion efficiency of 10%.With the Marcus model,in the BFBPD-PC61 BM interface,the exciton-dissociation rate,kdis,was predicted to be 2.684×10^13 s^-1,which is as 3-5 orders of magnitude large as the decay（radiative and non-radiative） one（10-8-10^10s^-1）,indicating a high exciton-dissociation efficiency of 100% in the BFBPD-PC61 BM interface.Furthermore,by the molecular dynamics simulation,the hole mobility of BFBPD thin-film was predicted to be as high as 1.265 × 10^-2 cm-2·V^-1·s^-1,which can be attributed to its dense packing in solid state.

Adaptive Graph Convolutional Recurrent Neural Networks for System-Level Mobile Traffic Forecasting

Accurate traffic pattern prediction in largescale networks is of great importance for intelligent system management and automatic resource allocation.System-level mobile traffic forecasting has significant challenges due to the tremendous temporal and spatial dynamics introduced by diverse Internet user behaviors and frequent traffic migration.Spatialtemporal graph modeling is an efficient approach for analyzing the spatial relations and temporal trends of mobile traffic in a large system.Previous research may not reflect the optimal dependency by ignoring inter-base station dependency or pre-determining the explicit geological distance as the interrelationship of base stations.To overcome the limitations of graph structure,this study proposes an adaptive graph convolutional network(AGCN)that captures the latent spatial dependency by developing self-adaptive dependency matrices and acquires temporal dependency using recurrent neural networks.Evaluated on two mobile network datasets,the experimental results demonstrate that this method outperforms other baselines and reduces the mean absolute error by 3.7%and 5.6%compared to time-series based approaches.

Dynamic data-sharing based user recruitment in mobile crowdsensing

Mobile crowdsensing(MCS) has become an emerging paradigm to solve urban sensing problems by leveraging the ubiquitous sensing capabilities of the crowd. One critical issue in MCS is how to recruit users to fulfill more sensing tasks with budget restriction, while sharing data among tasks can be a credible way to improve the efficiency. The data-sharing based user recruitment problem under budget constraint in a realistic scenario is studied, where multiple tasks require homogeneous data but have various spatio-temporal execution ranges, meanwhile users suffer from uncertain future positions. The problem is formulated in a manner of probability by predicting user mobility, then a dynamic user recruitment algorithm is proposed to solve it. In the algorithm a greedy-adding-and-substitution(GAS) heuristic is repeatedly implemented by updating user mobility prediction in each time slot to gradually achieve the final solution. Extensive simulations are conducted using a real-world taxi trace dataset, and the results demonstrate that the approach can fulfill more tasks than existing methods.

Mobility assisted spectrum aware routing protocol for cognitive radio ad hoc networks

We propose a mobility assisted spectrum aware routing(MASAR) protocol for cognitive radio ad hoc networks(CRAHNs),providing robustness to primary user activity and node mobility.This protocol allows nodes to collect spectrum information during a spectrum management interval followed by a transmission period.Cognitive users discover next hops based on the collected spectrum and mobility information.Using a beaconless mechanism,nodes obtain the mobility information and spectrum status of their neighbors.A geographical routing scheme is adopted to avoid performance degradation specially due to the mobility of the nodes and the activity of the primary users.Our scheme uses two approaches to fnd either short or stable routes.Since mobility metrics have a signifcant role in the selection of the next hop,both approaches use a reactive mobility update process assisted by mobility prediction to avoid location errors.MASAR protocol performance is investigated through simulations of diferent scenarios and compared with that of the most similar protocol,CAODV.The results indicate that MASAR can achieve signifcant reduction in control overhead as well as improved packet delivery in highly mobile networks.