45 entries: 22 journal articles, 20 conference papers, 1 book chapter, 2 patent applicationsGoogle Scholar
Showing 45 of 45
Martin Andersson, Tung T. Vu, Pål Frenger, Jan Åslund, Erik G. Larsson
Uses the fact that channels between access points change slowly to estimate them alongside the uplink data, suppressing the interference that downlink access points cause to uplink ones.
Pål Frenger, Tung T. Vu, Erik G. Larsson, Martin B. Andersson
Describes how an uplink access point can mitigate interference from another access point by measuring it during short phases when its own users are silent or send known samples.
Pål Frenger, Tung T. Vu, Erik G. Larsson, Martin B. Andersson
Describes how a device can suppress the interference another device causes in dynamic TDD networks, by estimating the channel between them and steering a beam null towards it.
Tung T. Vu, Swaroop Gopalam, Iain B. Collings, Stephen V. Hanly, Hazer Inaltekin
Schedules the beams of millimetre-wave base stations that reuse the full band so every user reaches its target rate in as few time slots as possible, and finds that purely analog beams outperform hybrid beamforming here.
Canh T. Dinh, Tung T. Vu, Nguyen H. Tran
Reviews the theory of personalised federated learning, in which each client keeps a model tailored to its own data, and sets out its open problems.
Unnikrishnan Kunnath Ganesan, Tung T. Vu, Erik G. Larsson
Lets cell-free access points cooperate in small phase-aligned clusters instead of across the whole network, reaching nearly the data rate of fully coherent operation without its strict phase alignment.
Chongzheng Hao, Tung T. Vu, Hien Q. Ngo, Minh N. Dao, Xiaoyu Dang, Chenghua Wang, Michail Matthaiou
Chooses which access points serve each user and at what power to maximise the total data rate, with a learning-based method (JointCFNet) for small networks and a fast gradient method for large ones.
Muhammad Farooq, Tung T. Vu, Hien Q. Ngo, Le-Nam Tran
Lets one massive MIMO base station serve federated-learning users and other users on the same time and frequency resources, keeping each learning round within its deadline while maximising the lowest rate among the other users.
Canh T. Dinh, Tung T. Vu, Nguyen H. Tran, Minh N. Dao, Hongyu Zhang
Treats federated learning on uneven client data as multi-task learning linked by a graph, giving one framework for standard and personalised federated learning and two algorithms (FedU, dFedU) with proven convergence rates.
Chi Y. Nguyen, Tung T. Vu, Hien Q. Ngo, Michail Matthaiou
Serves some cell-free users with non-orthogonal multiple access and the rest orthogonally under limited fronthaul, giving clearly higher total spectral efficiency than orthogonal access alone, most of all when coherence intervals are short.
Martin Andersson, Tung T. Vu, Pål Frenger, Erik G. Larsson
Mitigates the interference an uplink user causes to nearby downlink users in dynamic TDD by sending pilots between users, so each downlink user can design a combiner that rejects it.
Tung T. Vu, Swaroop Gopalam, Stephen V. Hanly, Iain B. Collings, Hazer Inaltekin
Assigns users to overlapping millimetre-wave base stations with limited RF chains so that all user data is cleared in the shortest time, using a two-stage scheme.
Martin Andersson, Tung T. Vu, Pål Frenger, Erik G. Larsson
Detects uplink symbols despite interference from a downlink access point in dynamic TDD, by keeping users silent for one sample per coherence interval to learn the interfering channel.
Mohammadali Mohammadi, Tung T. Vu, Hien Q. Ngo, Michail Matthaiou
Turns half-duplex access points into a virtual full-duplex cell-free network by assigning each to uplink or downlink, giving about 30% more spectral efficiency and up to 200% more energy efficiency than half- or full-duplex schemes.
Martin Andersson, Tung T. Vu, Pål Frenger, Erik G. Larsson
Adapts the uplink and downlink switching point of a cell-free network in dynamic TDD together with power control, to maximise energy efficiency.
Chongzheng Hao, Tung T. Vu, Hien Q. Ngo, Minh N. Dao, Xiaoyu Dang, Michail Matthaiou
Jointly chooses user association and power in user-centric cell-free massive MIMO with a fast accelerated gradient method, giving up to about 2.8 times the median spectral efficiency of a heuristic baseline.
Tung T. Vu, Hien Q. Ngo, Minh N. Dao, Erik G. Larsson
Uses network-assisted full-duplex in a cell-free relaying system, assigning access points to uplink or downlink, to raise per-pair spectral efficiency over half-duplex and heuristic baselines.
Tung T. Vu, Hien Q. Ngo, Minh N. Dao, Duy T. Ngo, Erik G. Larsson, Tho Le-Ngoc
Designs massive MIMO for federated learning in three ways (synchronous, asynchronous and session-based) to cut the energy the base station and users spend on each learning round without missing its deadline.
Tung T. Vu, Hien Q. Ngo, Minh N. Dao, Michail Matthaiou, Erik G. Larsson
Splits massive MIMO downlink transmission into sessions, each finishing one user so later sessions see less interference, to minimise the time needed to deliver all data.
Tung T. Vu, Duy T. Ngo, Hien Q. Ngo, Minh N. Dao, Nguyen H. Tran, Richard H. Middleton
Picks the users with the best links, and sets their power and processing speed, to cut the total time federated learning takes in cell-free massive MIMO while keeping enough users for good learning.
Muhammad Farooq, Tung T. Vu, Hien Q. Ngo, Le-Nam Tran
Serves federated-learning users in the uplink and other users in the downlink on two halves of the band, maximising the lowest rate of the other users while keeping learning rounds on time.
Mohammadali Mohammadi, Tung T. Vu, Behnaz Naderi Beni, Hien Q. Ngo, Michail Matthaiou
Assigns cell-free access points to uplink or downlink to create a virtually full-duplex network, which clearly beats half-duplex operation in total spectral efficiency.
Tung T. Vu, Trinh Van Chien, Canh T. Dinh, Hien Q. Ngo, Michail Matthaiou
Trains a neural network to estimate effective channel gains from received downlink signals in RIS-assisted massive MIMO, with no downlink pilots, where channel hardening is weak.
Tung T. Vu, Hien Q. Ngo, Duy T. Ngo, Minh N. Dao, Erik G. Larsson
Shows that one massive MIMO network can run several federated-learning processes at once, and allocates power and computing to save energy while keeping each iteration on time.
Tung T. Vu, Hien Q. Ngo, Thomas L. Marzetta, Michail Matthaiou
Asks how a wireless network can support several federated-learning groups at once, and answers with cell-free massive MIMO and a simple algorithm that minimises each iteration's time.
Tung T. Vu, Duy T. Ngo, Hien Q. Ngo, Minh N. Dao, Nguyen H. Tran, Richard H. Middleton
Tackles stragglers in wireless federated learning by choosing which users take part before training starts in cell-free massive MIMO, cutting transmission time.
Tung T. Vu, Duy T. Ngo, Nguyen H. Tran, Hien Q. Ngo, Minh N. Dao, Richard H. Middleton
Proposes cell-free massive MIMO as a stable network for federated learning, and jointly tunes local accuracy, power, data rate and processing speed to cut training time by up to 55% over baselines.
Tung T. Vu, Duy T. Ngo, Minh N. Dao, Quang-Thang Duong, Minoru Okada, Hung Nguyen-Le, Richard H. Middleton
Jointly designs precoding, power and rates for full-duplex cloud radio access networks to maximise energy efficiency, which beats half-duplex when self-interference is low.
Tung T. Vu, Duy T. Ngo, Hien Q. Ngo, Tho Le-Ngoc
Analyses full-duplex cell-free massive MIMO, shows that residual self-interference is its main limit, and finds a 140% spectral efficiency gain over half-duplex when that interference is low enough.
Thinh D. Tran, Long B. Le, Tung T. Vu, Duy T. Ngo
Models how wireless access and backhaul providers price and trade resources in network slicing as a Stackelberg game, and solves for its equilibrium.
Tung T. Vu, Duy T. Ngo, Minh N. Dao, Salman Durrani, Duy H. N. Nguyen, Richard H. Middleton
Maximises the energy efficiency of cloud radio access networks whose fronthaul links have limited capacity, and finds that sharing data over the fronthaul beats compressing it.
Tung T. Vu, Duy T. Ngo, Minh N. Dao, Salman Durrani, Richard H. Middleton
Decides which cache-equipped radio heads serve which users, and with what rates and precoding, in content-centric cloud radio access networks; edge caching improves both spectral and energy efficiency.
Tung T. Vu, Duy T. Ngo, Minh N. Dao, Salman Durrani, Duy H. N. Nguyen, Richard H. Middleton
Maximises the energy efficiency of a cloud radio access network whose data reaches the radio heads over capacity-limited fronthaul, with an algorithm that solves one simple convex problem per iteration.
Tung T. Vu, Ha H. Kha
Finds the globally optimal precoder for energy-efficient secure MIMO links with an eavesdropper as a benchmark, and a fast zero-forcing design that comes close to it.
Kha H. Ha, Tung T. Vu, Trung Q. Duong, Nguyen-Son Vo
Proposes two interference alignment schemes that protect multiuser MIMO links from an eavesdropper by aligning or nulling the signals it could intercept, raising the secrecy sum rate.
Tung T. Vu, Ha H. Kha, Trung Q. Duong, Nguyen-Son Vo
Uses particle swarm optimisation to design the precoders and receivers of a multiuser MIMO broadcast channel that maximise the weighted sum rate.
Ha H. Kha, Tung T. Vu, Tuan Do-Hong
Designs energy-efficient transceivers for multiuser MIMO cognitive radio, where primary users align out their mutual interference and secondary users avoid interfering with them.
Tung T. Vu, Ha H. Kha, Osamu Muta, Mohamed Rihan
Coordinates macro and small cells in two-tier MIMO heterogeneous networks with little channel information exchange, using interference alignment and power allocation to maximise energy efficiency.
Tung T. Vu, Duy T. Ngo, Lawrence Ong, Salman Durrani, Richard H. Middleton
Jointly designs user association, delivery rates and precoding in cache-enabled fog radio access networks, raising throughput by 50% while saving at least half the power of precoding alone.
Tung T. Vu, Ha H. Kha, Hoang D. Tuan
Uses interference alignment to block information leakage in secure multiuser MIMO systems, then maximises energy efficiency with a semi-closed-form algorithm.
Tran T. Duy, George C. Alexandropoulos, Tung T. Vu, Vo N. Son, Trung Q. Duong
Derives the outage probability of vehicle-to-vehicle links relayed through roadside units in cognitive radio, under double Rayleigh fading and several relay selection policies.
Tung T. Vu, Ha H. Kha
Benchmarks secure, energy-efficient MIMO precoding with a globally optimal method, and proposes a fast zero-forcing design that cancels the eavesdropper's signal.
Dac-Binh Ha, Tung T. Vu, Tran T. Duy, Vo N. Q. Bao
Analyses cognitive radio relay networks with decode-and-forward relays, deriving the outage probability and, when an eavesdropper is present, the average secrecy capacity.
Tung T. Vu, Ha H. Kha, Trung Q. Duong
Casts secure interference alignment as rank minimisation, aligning interference and eavesdropped signals into the smallest subspaces, which raises the secrecy sum rate.
Tung T. Vu, Ha H. Kha, Trung Q. Duong, Nguyen-Son Vo
Designs interference alignment that minimises both interference and the signals an eavesdropper can intercept in multiuser MIMO, improving the secrecy sum rate.