Associate Professor Duy Ngo

Considering a dense small-cell network with simultaneous wireless information and power transfer (SWIPT), this work jointly designs transmit beamformers at the base stations (BSs) and receive power splitting ratios at the users (UEs). Our objectives is to maximize the minimum UE signal-to-interference-plus-noise-ratio (SINR) under BS transmit power and UE minimum harvested energy constraints. This problem is highly nonconvex, for which semidefinite programming (SDP) relaxation may even fail to locate a feasible solution. We propose an efficient spectral optimization method by expressing the rank-one constraints as a single reverse convex nonsmooth constraint and incorporating it in the optimization objective. The proposed algorithm practically achieves the theoretical bound given by SDP relaxation with almost similar complexity.

Cooperative communications using the Alamouti space-time block code (STBC) has been mentioned intensively. However, to our knowledge, an in-depth error performance analysis for a space-time coded cooperative communications system, where each node acts as both a source node and a relay node for its partner, has not been reported. Intuitively, cooperative communications would not always be better than direct transmission, but only under certain conditions. This paper first derives an exact performance analysis of the Alamouti STBC cooperative communications for a Binary Phase Shift Keying (BPSK) constellation, which could be extended to a higher density modulation scheme. This mathematical analysis facilitates the source nodes to decide pro-actively whether cooperative communications or direct transmission should be used, depending on the channel conditions between the sources themselves and the channel conditions between the sources and the destination. The anterior knowledge of which transmission mode should be used before nodes actually engage in cooperation is useful to keep system operations as simple as possible, while assuring that cooperative communications is definitely beneficial once it is in place. The analytical error performances are then verified by simulations and in-depth discussions on the simulation results are presented.

Cognitive networks (CN) introduce an adaptable network architecture which enables efficient sharing of transmission resources among heterogeneous networks. Traditional CN architecture requires each network devices to independently employ channel sensing mechanisms to discover free (or white) spectrum that can be temporarily borrowed for its own use. However, the distributed physical layer-based channel borrowing techniques cannot guarantee the quality-of-service (QoS) requirements of primary and secondary users in a CN due to the lack of channel availability information. On the other hand, the FCC regulations introduced in 2010 allowed the use of Geo-Location Database (GDB) as the primary means of determining white space availability. Using GDB-based CN network architecture could offer significant QoS advantages. In this paper, we propose a new GDB-based QoS-aware CN architecture and secondary channel allocation technique that guarantees predictable and stable QoS for the secondary users. The proposed channel allocation technique uses statistical QoS estimation to predict the radio resource requirements of secondary users. We also propose to deploy a centralized channel manager that uses both proactive and reactive resource allocation techniques to allocate the white spectrum among secondary users. An OMNeT++ based CN simulation model has been developed to evaluate the proposed secondary channel allocation technique for an IEEE802.11 based cognitive WiFi network. Simulation results show that the proposed secondary channel allocation technique offers significant QoS advantages and stability to secondary users compared to the conventional distributed channel sensing technique.

This paper is concerned with the optimal association of user equipments (UEs) to the base stations, which are either a macro base station or pico base stations, for the sum rate utility in a cooperative heterogeneous network (HetNet). This is a nonconvex combinatorial problem. By recasting the problem as a d.c (difference of convex (d.c.)) program, the optimal association solution is sought by d.c iterations (DCI), each of which solves a simple convex quadratic problem over a convex set. Numerical simulation shows the efficiency of the proposed procedure in term of performance and computational time.

We consider a multicell network where an amplify-and-forward relay is deployed in each cell to help the base station (BS) serve its cell-edge user. We assume that each relay scavenges energy from all received radio signals to process and forward the information data from the BS to the corresponding user. For this, a power splitter and a wireless energy harvester are implemented in the relay. Our aim is to minimize the total power consumption in the network while guaranteeing minimum data throughput for each user. To this end, we develop a resource management scheme that jointly optimizes three parameters, namely, BS transmit powers, power splitting factors for energy harvesting and information processing at the relays, and relay transmit powers. As the formulated problem is highly nonconvex, we devise a successive convex approximation algorithm based on difference-of-convex-functions (DC) programming. The proposed iterative algorithm transforms the nonconvex problem into a sequence of convex problems, each of which is solved efficiently in each iteration. We prove that this path-following algorithm converges to an optimal solution that satisfies the Karush-Kuhn-Tucker (KKT) conditions of the original nonconvex problem. Simulation results demonstrate that the proposed joint optimization solution substantially improves the network performance.

With the increasing deployment of wireless sensor networks (WSNs), the demand for energy efficiency (EE) and Quality of Service (QoS) guarantee for such networks is on the rise. In a distributed WSN, the EE could be directly related to the operational expenditure (OPEX) due to the frequent replacement of batteries. An increasing EE helps prolong the network lifetime. Energy efficiency in a WSN can be improved by optimising PHY and MAC layer algorithms. In this paper, we introduce an adaptive energy efficiency algorithm, known as ABSD that changes the MAC parameters of the IEEE 802.15.4 sensor nodes in response to the queue occupancy level of sensor nodes and the offered traffic load conditions. By adapting the transmission parameters, the ABSD algorithm minimises the network contention which could in turn improve the energy efficiency as well as the throughput. The paper also presents a mathematical model to characterise the energy consumption processes controlled by the IEEE 802.15.4 MAC layer. Our ABSD algorithm is simulated by using an OPNET-based discrete event simulation model. Simulation results show that the proposed algorithm offers high EE and maintains high QoS for different offered load conditions.

This paper considers a heterogeneous multicell network where the base station (BS) in each cell communicates with its cell-edge user with the assistance of an amplify-and-forward relay node. Equipped with a power splitter and a wireless energy harvester, the relay scavenges RF energy from the received signals to process and forward the information. In the face of strong intercell interference and limited radio resources, we develop a resource allocation scheme that jointly optimizes (i) BS transmit powers, (ii) power splitting factors for energy harvesting and information processing at the relays, and (iii) relay transmit powers. To solve the highly non-convex problem formulation of sum-rate maximization, we propose to apply the successive convex approximation (SCA) approach and devise an iterative algorithm based on geometric programming. The proposed algorithm transforms the nonconvex problem into a sequence of convex problems, each of which is solved very efficiently by the interior-point method. We prove that our developed algorithm converges to an optimal solution that satisfies the Karush-Kuhn-Tucker conditions of the original nonconvex problem. Numerical results confirm that our joint optimization solution substantially improves the network performance, compared to the existing solution wherein only the received power splitting factors at the relays are optimized.

This paper develops power allocation schemes to manage the signal interference in a multiuser multicell network, where full-duplex transceivers are implemented at all base stations (BSs) and user equipment units (UEs). Beside the acute intracell and intercell interferences, the significant residual self-interference at the full-duplex receivers is the limiting factor for any network performance enhancement. To help control such severe interferences, we propose to associate each bidirectional full-duplex link with a net utility function, which consists of a utility and a flexible price. While the utility corresponds to the link throughput, the proposed logarithmic cost function allows for a moderate penalty. Our aim is to maximize the sum network utility subject to the power constraints at the BSs and UEs. To solve the highly nonconvex problem formulation, we propose two successive convex approximation (SCA) algorithms based on the difference-of-convex-functions programming and the geometric programming. In each algorithm, we specifically tailor the generic SCA framework and transform our formulated nonconvex problem into a sequence of convex power allocation programs. We prove that the developed iterative algorithms converge to locally optimal solutions that satisfy the Karush-Kuhn-Tucker conditions of the original problem. Numerical results confirm that our utility-based solutions markedly improve the network throughput by effectively managing the interferences.

This paper considers the Han-Kobayashi transmission strategy that mitigates the downlink intercell interference in a multiuser multi-input multi-output (MU-MIMO) multicell network. The base station (BS) splits the transmitted data of a user (UE) into a common message and a private message, the former of which is then decoded by a paired UE in another cell so as to reduce the respective cross-cell interference. Our aim here is to develop (i) A pairing rule that determines pairs of UEs to share common messages, and (ii) Optimal precoders at the BSs that maximize either the minimum UE throughput or the network sum-rate. To solve the combinatoric UE pairing problem, we propose a heuristic that pairs UEs with the largest corresponding cross-cell channel gains. This approach ensures that the most significant source of intercell interference is eliminated through common message decoding. We then apply the Frank-and-Wolfe procedure of concave programming to solve the highly nonconvex precoder design problems. We show that this procedure generates a sequence of improved solutions and eventually converges to at least a local optimum. Numerical results confirm the advantages of our proposed solution over the conventional strategy where intercell interference is treated as noise.

In a vehicular ad hoc network (VANET), basic safety messages (BSM) are exchanged among vehicles to provide cooperative awareness and support safety applications. While the periodic generation and transmission of such messages already place a heavy burden on the network load, the hidden nodes further complicate the network situation with packet collisions being undetectable. In this paper, we propose a space-division multiple access (SDMA) technique combined with an adaptive rate control mechanism to improve the efficiency of BSM transmissions. With SDMA to reduce the interference among the vehicles and address the hidden-node problem, only vehicles located sufficiently far apart are allowed to reuse the same time slot to send their BSMs. The developed SDMA approach permits multiple transmissions per road segment and uses Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) in each segment, giving rise to a much better use of both space and bandwidth. To guarantee that the actual number of BSM transmissions does not exceed the maximum allowable, we devise an adaptive scheme that optimally adjusts the individual packet generation rates in accordance to the vehicle density. Toward this end, we find an optimal bandwidth utilization that maximizes the newly-defined metric, termed as 'safety throughput' per vehicle. Such an optimal utilization allows a large number of BSM packets to be generated while maintaining a high reception probability. Our comprehensive simulation results confirm the clear advantages of our proposal over the existing techniques, in terms of safety range, BSM packet inter-arrival time, and safety throughput.

The problems of designing signature sequences and power allocation policy for code-division multiple access (CDMA) are important and have been the subject of intensive research in recent years. Two different criteria adopted in such design problems are the user capacity and the information-theoretic capacity. Regarding the maximization of the information-theoretic capacity, most of the previous works only consider the optimizations of signature sequences and power allocation separately. In contrast, this paper presents a jointly optimal design of signature sequences and power allocation under the sum power constraint. The proposed design is of closed-form and applicable for the general case of correlated signals and colored noise. Numerical results verify the superiority of the proposed design over the existing ones. © 2007 IEEE.

Unmanned aerial vehicles (UAVs)-assisted wireless sensor networks (WSNs) have tremendous potential applications due to their flexibility and rapid deployment. Hybrid time-of-arrival (TOA) and angle-of-arrival (AOA) localization techniques are commonly used due to their high accuracy. The conventional TOA/AOA localization algorithms require both zenith and azimuth angle estimations at each agent. Since a zenith-and-azimuth AOA estimation requires an L-shape antenna array and complicated 2-D signal processing, conventional TOA/AOA algorithms lead to both hardware and computational complexity and high power consumption at each agent. This letter proposes a hybrid TOA/AOA localization algorithm, named $\text{T1A}_{a}$, to reduce the agents' complexity. $\text{T1A}_{a}$ only combines TOA-ranging with the azimuth angle estimation. Thereby, agents only require a 1-D antenna array, less complicated signal processing, and thus, lower power consumption. Such improvements are important for power-limited WSNs and Internet-of-Things (IoT) systems. Simulations are conducted to prove that the proposed $\text{T1A}_{a}$ technique can achieve similar performance as conventional TOA/AOA methods, while significantly simplifying agents' complexity.

This work proposes novel synchronous, asynchronous, and session-based designs for energy-efficient massive multiple-input multiple-output networks to support federated learning (FL). The synchronous design relies on strict synchronization among users when executing each FL communication round, while the asynchronous design allows more flexibility for users to save energy by using lower computing frequencies. The session-based design splits the downlink and uplink phases in each FL communication round into separate sessions. In this design, we assign users such that one of the participating users in each session finishes its transmission and does not join the next session. As such, more power and degrees of freedom will be allocated to unfinished users, resulting in higher rates, lower transmission times, and hence, higher energy efficiency. In all three designs, we use zero-forcing processing for both uplink and downlink, and develop algorithms that optimize user assignment, time allocation, power, and computing frequencies to minimize the energy consumption at the base station and users, while guaranteeing a predefined maximum execution time of each FL communication round.

Automotive infotainment systems are expected to be first deployed on highways to service drivers travelling long distances, who are more likely to utilize the infotainment applications. In order to meet the stringent requirements of the infotainment systems, road side units (RSUs) are installed along the highway to facilitate a continuous vehicle-to-infrastructure (V2I) connectivity. Due to the long travelling distance and small coverage of the individual RSU, a more cost-effective solution would be to combine V2I with the vehicle-to-vehicle (V2V) communications to maintain the continuous connectivity. In this paper, we propose a new dynamic cooperation scheme that employs a dynamic forwarder selection strategy to generate an adaptive multi-hop V2V path for connectivity maintenance and throughput enhancement at a vehicle located outside of the RSU's coverage range. For the commonly assumed scenario that all vehicles travel in the same direction and at the same speed, we develop an analytical model and derive closed-formed expressions for the average out-of-range connection time, number of service resumptions and achieved throughput. The developed analytical model provides insights into the impacts of inter-RSU distance, vehicles' assistance willingness and the target vehicle's buffer size to the network performance. Simulation results with practical parameter settings show that our proposed scheme is effective in improving connectivity while offering a high throughput for the target vehicle. In particular, a high vehicle density, more assistance willingness by the forwarders and a large buffer size at the target vehicle are shown to be helpful in sparse RSU deployments.

Infrastructure-to-vehicle (I2V) and vehicle-to-vehicle (V2V) communications are often combined to extend the connectivity and coverage in the Intelligent Transportation System (ITS) and its applications, e.g., augmented reality, real-time parking management and online shopping. Through multi-hop I2V and V2V communications, requesting vehicles are always connected to road side units (RSUs) even when they do not reside within the RSUs' coverage range. However, there may be not adequate network resource for several I2V and V2V links when multiple vehicles request services simultaneously. In this paper, we propose a joint frequency scheduling and power control scheme to enhance connectivity in multi-hop I2V/V2V networks. We associate I2V and V2V links with tuple-links, then formulate an NP-hard problem in which a frequency scheduler and a power controller are jointly designed for the tuple-links. The NP-hard problem is decomposed into two separate subproblems by employing the delayed column generation technique. Then, we employ a method for linear programming and a greedy algorithm to address these subproblems. Through numerical experiments with practical parameter settings, we demonstrate the proposed scheme outperforms several existing ones in terms of connectivity enhancement, measured by the service resumption number and average achieved throughput. Furthermore, the efficiency of our scheme is further enhanced when the number of available channels is high, and buffer size equipped to the requesting vehicles is large.

This paper addresses the tradeoff problem between hit ratio and content quality in edge caching systems for multiuser adaptive bitrate streaming (ABS) services. A dynamic policy for cache decision and quality level selection for each ABS content during every cache cycle is proposed. Achieving this policy is NP-complete. For this, the considered problem is transformed into a nested multidimensional 0/1 knapsack optimization problem which is then resolved by a cooperative transfer learning-accelerated genetic algorithm. Performance evaluation demonstrates an adaptation of the proposed algorithm on various video stream popularity models in terms of algorithmic convergence and cache balancing.

This article considers a dual-hop Internet of Things communications network where sensor nodes transmit data to a gateway either directly or via other nodes using dual-hop communications. Each node employs separate transmission buffers to store its own sensing data and data received from other nodes. End-to-end delay quality-of-service constraints in terms of the maximum acceptable delay-outage probabilities are imposed. We investigate energy-efficient adaptive resource allocation problems (i.e., joint link scheduling, rate, and power allocation) to support minimum data rates of the nodes. A novel approach is proposed exploiting asymptotic delay analysis to first determine the achieved delay exponents of the queue length tail distributions to satisfy the delay-outage constraints. Next, the relation between the delay exponents and resource allocation variables are derived. Last, the solutions to the resulting constrained optimization problems are obtained using the Lagrangian approach and convex optimization. Illustrative examples demonstrate the effects of the rate requirements and delay constraint stringency on the power consumption and routing configuration.

In this paper, new strategies are devised for joint load balancing and interference management in the downlink of a heterogeneous network, where small cells are densely deployed within the coverage area of a traditional macrocell. Unlike existing work, the limited backhaul capacity at each base station (BS) is taken into account. Here, users (UEs) cannot be offloaded to any arbitrary BS, but only to ones with sufficient backhaul capacity remaining. Jointly designed with traffic offload, transmit power allocation mitigates the intercell interference to further support the quality of service of each UE. The objective here is either: 1) to maximize the network sum rate subject to minimum throughput requirements at individual UEs, or 2) to maximize the minimum UE throughput. Both formulated problems belong to the difficult class of mixed-integer nonconvex optimization problems. The inherently binary BS-UE association variables are strongly coupled with the transmit power variables, making the problems even more challenging to solve. New iterative algorithms are developed based on an exact penalty method combined with successive convex programming, where the binary BS-UE association problem and the nonconvex power allocation problem are dealt with one at a time. At each iteration of the proposed algorithms, only two simple convex problems need to be solved at the same time scale. It is proven that the algorithms improve the objective functions at each iteration and converge eventually. Numerical results demonstrate the efficiency of the proposed algorithms in both traffic offloading and interference mitigation.

For a multiuser multi-input multi-output (MU-MIMO) multicell network, the Han-Kobayashi strategy aims to improve the achievable rate region by splitting the data information intended to a serviced user (UE) into a common message and a private message. The common message is decodable by this UE and another UE from an adjacent cell so that the corresponding intercell interference is cancelled off. This work aims to design optimal precoders for both common and private messages to maximize the network sum-rate, which is a highly nonlinear and nonsmooth function in the precoder matrix variables. Existing approaches are unable to address this difficult problem. In this paper, we develop a successive convex quadratic programming algorithm that generates a sequence of improved points. We prove that the proposed algorithm converges to at least a local optimum of the considered problem. Numerical results confirm the advantages of our proposed algorithm over conventional coordinated precoding approaches where the intercell interference is treated as noise.

Information and energy can be transferred over the same radio-frequency channel. In the power-splitting (PS) mode, they are simultaneously transmitted using the same signal by the base station (BS) and later separated at the user (UE)'s receiver by a power splitter. In the time-switching (TS) mode, they are either transmitted separately in time by the BS or received separately in time by the UE. In this paper, the BS transmit beamformers are jointly designed with either the receive PS ratios or the transmit TS ratios in a multicell network that implements wireless information and power transfer (WIPT). Imposing UE-harvested energy constraints, the design objectives include: 1) maximizing the minimum UE rate under the BS transmit power constraint, and 2) minimizing the maximum BS transmit power under the UE data rate constraint. New iterative algorithms of low computational complexity are proposed to efficiently solve the formulated difficult nonconvex optimization problems, where each iteration either solves one simple convex quadratic program or one simple second-order-cone-program. Simulation results show that these algorithms converge quickly after only a few iterations. Notably, the transmit TS-based WIPT system is not only more easily implemented but outperforms the receive PS-based WIPT system as it better exploits the beamforming design at the transmitter side.

We consider the joint design of transmit beamforming and receive signal-splitting ratios in the downlink of a wireless network with simultaneous radio frequency information and energy transfer. Under constraints on the signal-to-interference-plus-noise ratio at each user and the total transmit power at the base station, the design objective is to maximize either the sum harvested energy or the minimum harvested energy. We develop a computationally efficient path-following method to solve these challenging nonconvex optimization problems. We mathematically show that the proposed algorithms iteratively progress and converge to locally optimal solutions. Simulation results further show that these locally optimal solutions are the same as the globally optimal solutions for the considered practical network settings.

This paper designs jointly optimal linear precoders for both base stations (BSs) and users in a multiuser multi-input multi-output (MU-MIMO) multicell network. The BSs are full-duplexing transceivers while uplink users and downlink users (DLUs) are equipped with multiple antennas. Here, the network quality-of-service (QoS) requirement is expressed in terms of the minimum throughput at the BSs and DLUs. We consider the problems of either QoS-constrained sum throughput maximization or minimum cell throughput maximization. Due to the nonconcavity of the throughput functions, the optimal solutions of these two problems remain unknown in both half-duplexing and full-duplexing networks. The first problem has a nonconcave objective and a nonconvex feasible set, whereas the second problem has a nonconcave and nonsmooth objective. To solve such challenging optimization problems, we develop iterative low-complexity algorithms that only invoke one simple convex quadratic program at each iteration. Since the objective value is proved to iteratively increase, our path-following algorithms converge at least to the local optimum of the original nonconvex problems. Due to their guaranteed convergence, simple implementation, and low complexity, the devised algorithms lend themselves to practical precoder designs for large-scale full-duplex MU-MIMO multicell networks. Numerical results demonstrate the advantages of our successive convex quadratic programming framework over existing solutions.

The periodic generation and transmission of basic safety messages (BSM) place a heavy burden on the load of a vehicular ad hoc network (VANET), where there also remain multiple packet collisions undetected due to the hidden terminals. In this paper, we propose a distributed scheme that controls the spatial reuse distance to improve the efficiency of BSM transmissions in space-division multiple access (SDMA) based VANETs. With the proposed SDMA structure, only vehicles located sufficiently far apart are allowed to reuse the same time slot to send their BSMs. The signal interference is thus reduced while the hidden-node problem is essentially alleviated. Multiple transmissions per SDMA road segment or 'cell' are also enabled using the Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA), giving rise to a much better use of both space and bandwidth. To guarantee that the actual number of BSM transmissions does not exceed the maximum allowable in each SDMA cell, we further devise an adaptive scheme that adjusts the spatial reuse distance in accordance with the vehicle density. Because the global information of vehicle density is not available at individual vehicles, we propose a distributed algorithm that estimates the vehicle density and makes consensus to enhance the accuracy of spatial reuse distance estimation. As the transmission range is controlled accordingly, the mutual interference among the SDMA cells is further reduced. Importantly, the developed control algorithm can be distributively implemented by each vehicle with limited information exchange. To optimize the performance of the proposed solution, we also determine the optimal bandwidth utilization that maximizes the newly-defined criterion termed as 'safe range,' an important figure-of-merit in vehicular safety applications. Simulation results confirm the clear advantages of our proposal over the available approaches in terms of safety range, packet reception rate, end-to-end delay and BSM inter-arrival time in realistic network scenarios.

In this paper, we propose a joint subchannel and power allocation algorithm for the downlink of an orthogonal frequency-division multiple access (OFDMA) mixed femtocell/macrocell network deployment. Specifically, the total throughput of all femtocell user equipments (FUEs) is maximized while the network capacity of an existing macrocell is always protected. Towards this end, we employ an iterative approach in which OFDM subchannels and transmit powers of base stations (BS) are alternatively assigned and optimized at every step. For a fixed power allocation, we prove that the optimal policy in each cell is to give each subchannel to the user with the highest signal-to-interference-plus-noise ratio (SINR) on that subchannel. For a given subchannel assignment, we adopt the successive convex approximation (SCA) approach and transform the highly nonconvex power allocation problem into a sequence of convex subproblems. In the arithmetic-geometric mean (AGM) approximation, we apply geometric programming to find optimal solutions after condensing a posynomial into a monomial. On the other hand, logarithmic and difference-of-two-concave-functions (D.C.) approximations lead us to solving a series of convex relaxation programs. With the three proposed SCA-based power optimization solutions, we show that the overall joint subchannel and power allocation algorithm converges to some local maximum of the original design problem. While a central processing unit is required to implement the AGM approximation-based solution, each BS locally computes the optimal subchannel and power allocation for its own servicing cell in the logarithmic and D.C. approximation-based solutions. Numerical examples confirm the merits of the proposed algorithm. © 2014 IEEE.