Generalized Global Bandit and Its Application in Cellular Coverage Optimization

Shen, Cong; Zhou, Ruida; Tekin, Cem; Schaar, Mihaela van der

doi:10.1109/jstsp.2018.2798164

Cited by 29 publications

(20 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…MAB theory has been widely applied in wireless networks, such as power allocation in small base stations [16] [17], content placement in edge caching [18], [19], task assignment in mobile crowdsourcing [20] and mobility management in mobile edge computing [21]. Very recently, the BA problem is studied based on MAB theory, which makes online decision to strike the balance between exploitation and exploration.…”

Section: Related Workmentioning

confidence: 99%

“…Furthermore, the weak Lipschitz condition is mild, which only has the impact on the region in the vicinity of the optimal value. This assumption is well justified in many practical applications [17]. 2) (Well-shaped region) For a region, C h,j , of depth h, the region contains a ball with a radius of ρ 2 γ h which locates in the center of C h,j .…”

Section: Regret Performance Analysismentioning

confidence: 99%

See 1 more Smart Citation

Fast mmwave Beam Alignment via Correlated Bandit Learning

Cheng

Zhang

et al. 2019

IEEE Trans. Wireless Commun.

128

View full text Add to dashboard Cite

Beam alignment (BA) is to ensure the transmitter and receiver beams are accurately aligned to establish a reliable communication link in millimeter-wave (mmwave) systems. Existing BA methods search the entire beam space to identify the optimal transmit-receive beam pair, which incurs significant BA latency on the order of seconds in the worst case. In this paper, we develop a learning algorithm to reduce BA latency, namely Hierarchical Beam Alignment (HBA) algorithm. We first formulate the BA problem as a stochastic multi-armed bandit problem with the objective to maximize the cumulative received signal strength within a certain period. The proposed algorithm takes advantage of the correlation structure among beams such that the information from nearby beams is extracted to identify the optimal beam, instead of searching the entire beam space. Furthermore, the prior knowledge on the channel fluctuation is incorporated in the proposed algorithm to further accelerate the BA process.Theoretical analysis indicates that the proposed algorithm is asymptotically optimal. Extensive simulation results demonstrate that the proposed algorithm can identify the optimal beam with a high probability and reduce the BA latency from hundreds of milliseconds to a few milliseconds in the multipath channel, as compared to the existing BA method in IEEE 802.11ad.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Regret Performance Analysismentioning

confidence: 99%

Fast mmwave Beam Alignment via Correlated Bandit Learning

Cheng

Zhang

et al. 2019

IEEE Trans. Wireless Commun.

128

View full text Add to dashboard Cite

show abstract

“…Dandanov et al [14] proposed an RL method to optimize the network performance and discussed the effect of antenna down tilts on network coverage and capacity. Cong et al [15] presented a generalized global bandit method to solve the cellular coverage optimization problem.…”

Section: Related Workmentioning

confidence: 99%

“…Learning-based methods tend to solve the optimization problem by learning necessary information from available data or by interactions with the surrounding environments and make suitable decisions on future action based on some learned models or in a model-free manner [14], [15].…”

Section: Introductionmentioning

confidence: 99%

Accelerated Coverage Optimization With Particle Swarm in the Quotient Space Characterizing Antenna Azimuths of Cellular Networks

et al. 2019

View full text Add to dashboard Cite

In order to provide a reliable connection for the ever-increasing devices, cellular networks are facing critical challenges, among which is the issue of network coverage optimization. The costly objective function for network coverage appeals for an efficient approach of optimization, as the canonical particle swarm optimization (PSO) suffers excessive computation caused by population and iteration. The specificity of the antenna azimuths in cellular networks is hereby investigated so as to construct the corresponding metric structures in the naive antenna azimuth space and its quotient space. We introduce two accelerated PSO algorithms induced by the metric structures to maximize the cellular network coverage ratio. The former, named aPSO/S, only considers the shortest arc distance in the naive space, whereas the latter, named aPSO/QS, considers both the shortest arc distance and the equivalence among solutions in the quotient space. The proposed PSO algorithms, mainly the latter as referred to, not only guide the particles to fly toward the shortest arc distance directions but also eliminate lots of unnecessary calculation caused by the equivalence among solutions. The experiments show that the proposed aPSO/QS algorithm edge outs the canonical algorithm and its typical derivatives with various boundary conditions from the perspective of either the convergence speed or the stability with the hyper-parameters. To reach the same performance as that of its best competitor, the proposed aPSO/QS algorithm needs only a fraction of the computation cost so as to be applied in the efficiency-aware live network optimization.

show abstract

“…The structured bandit setting considered in this paper can be viewed as a hidden context problem, where the objective is to do targeted advertising for a user without observing their features, as illustrated in Figure 1. Apart from ad selection, the structured bandit model also has applications in dynamic pricing (described in [7]), cellular coverage optimization (by [8]), drug dosage optimization (discussed in [9]).…”

Section: Introductionmentioning

confidence: 99%

A Unified Approach to Translate Classical Bandit Algorithms to Structured Bandits

Gupta

Chaudhari

Mukherjee

et al. 2021

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

We consider a finite-armed structured bandit problem in which mean rewards of different arms are known functions of a common hidden parameter θ * . This problem setting subsumes several previously studied frameworks that assume linear or invertible reward functions. We propose a novel approach to gradually estimate the hidden θ * and use the estimate together with the mean reward functions to substantially reduce exploration of sub-optimal arms. This approach enables us to fundamentally generalize any classic bandit algorithm including UCB and Thompson Sampling to the structured bandit setting. We prove via regret analysis that our proposed UCB-C and TS-C algorithms (structured bandit versions of UCB and Thompson Sampling, respectively) pull only a subset of the sub-optimal arms O(log T ) times while the other sub-optimal arms (referred to as non-competitive arms) are pulled O(1) times. As a result, in cases where all sub-optimal arms are noncompetitive, which can happen in many practical scenarios, the proposed algorithms achieve bounded regret.

show abstract

Generalized Global Bandit and Its Application in Cellular Coverage Optimization

Cited by 29 publications

References 33 publications

Fast mmwave Beam Alignment via Correlated Bandit Learning

Fast mmwave Beam Alignment via Correlated Bandit Learning

Accelerated Coverage Optimization With Particle Swarm in the Quotient Space Characterizing Antenna Azimuths of Cellular Networks

A Unified Approach to Translate Classical Bandit Algorithms to Structured Bandits

Contact Info

Product

Resources

About