Constrained Online Convex Optimization With Feedback Delays

Cao, Xuanyu; Zhang, Junshan; Poor, H. Vincent

doi:10.1109/tac.2020.3030743

Cited by 22 publications

(16 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The former centered on online gradient descent in the unconstrained case while the latter was based on online mirror descent and FTRL-prox. Very recently, Cao et al (2020) extended the delayed feedback analysis to an online saddle-point algorithm which handled the constraints through Lagrangian relaxation.…”

Section: Related Workmentioning

confidence: 99%

Multi-Agent Online Optimization with Delays: Asynchronicity, Adaptivity, and Optimism

Hsieh¹,

Iutzeler²,

Malick³

et al. 2020

Preprint

View full text Add to dashboard Cite

Online learning has been successfully applied to many problems in which data are revealed over time. In this paper, we provide a general framework for studying multi-agent online learning problems in the presence of delays and asynchronicities. Specifically, we propose and analyze a class of adaptive dual averaging schemes in which agents only need to accumulate gradient feedback received from the whole system, without requiring any between-agent coordination. In the single-agent case, the adaptivity of the proposed method allows us to extend a range of existing results to problems with potentially unbounded delays between playing an action and receiving the corresponding feedback. In the multi-agent case, the situation is significantly more complicated because agents may not have access to a global clock to use as a reference point; to overcome this, we focus on the information that is available for producing each prediction rather than the actual delay associated with each feedback. This allows us to derive adaptive learning strategies with optimal regret bounds, at both the agent and network levels. Finally, we also analyze an "optimistic" variant of the proposed algorithm which is capable of exploiting the predictability of problems with a slower variation and leads to improved regret bounds.

show abstract

Section: Related Workmentioning

confidence: 99%

Multi-Agent Online Optimization with Delays: Asynchronicity, Adaptivity, and Optimism

Hsieh¹,

Iutzeler²,

Malick³

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…This demonstrates the effectiveness of the proposed spatial isolation approach via simultaneously sharing all frequency resources among SPs. We observe that in a wide range of Pc and N c values, our proposed precoding solution substantially outperforms the online solution from [20], the offline benchmark, and the FD ZF scheme.…”

Section: Simulation Resultsmentioning

confidence: 91%

“…), where σ 2 n = N 0 B W + N F , and [A] i,j denotes the (i, j) element of matrix A. For performance comparison, we consider 1) the online algorithm in [20], which is currently the best constrained OCO algorithm that accommodates multi-slot feedback delay; 2) the standard offline benchmark V ⋆ ; 3) a frequency division (FD) scheme that allocates equal bandwidth BW M to each SP m. This FD approach is commonly adopted in the existing literature of MIMO WNV [2]- [4]. At each cell c, each SP m adopts ZF precoding to serve its K m c users directly based on the local CSI.…”

Section: Simulation Resultsmentioning

confidence: 99%

“…2 shows f (T ) and P (T ) versus T with different values of CSI feedback delay τ . Our online precoding solution outperforms the one in [20], which uses the gradient ∇g(V) information to minimize g(V). In our solution, we directly minimize the constraint function g(V) as one part of the objective function in P2, which improves the control of the transmit power.…”

Section: Simulation Resultsmentioning

confidence: 99%

“…In our solution, we directly minimize the constraint function g(V) as one part of the objective function in P2, which improves the control of the transmit power. Furthermore, the regularization is imposed on both V t−τ and V t−1 instead of the single regularization term in [20]. The double regularization improves the efficiency of gradient descent on minimizing the precoding deviation.…”

Section: Simulation Resultsmentioning

confidence: 99%

See 2 more Smart Citations

Online Distributed Coordinated Precoding for Virtualized MIMO Networks with Delayed CSI

Wang,

Liang,

Dong

et al. 2022

Preprint

View full text Add to dashboard Cite

We consider online wireless network virtualization (WNV) in a multi-cell multiple-input multiple output (MIMO) system with delayed feedback of channel state information (CSI). Multiple service providers (SPs) simultaneously share the base station resources of an infrastructure provider (InP). We aim at minimizing the accumulated precoding deviation of the InP's actual precoder from the SPs' virtualization demands via managing both inter-SP and inter-cell interference, subject to both long-term and short-term per-cell transmit power constraints. We develop an online coordinated precoding solution and show that it provides provable performance bounds. Our precoding solution is fully distributed at each cell, based only on delayed local CSI. Furthermore, it has a closed-form expression with low computational complexity. Finally, simulation results demonstrate the substantial performance gain of our precoding solution over the current best alternative.

show abstract

Recent advances on distributed online optimization

2021

Control Theory Technol.

View full text Add to dashboard Cite

Constrained Online Convex Optimization With Feedback Delays

Cited by 22 publications

References 8 publications

Multi-Agent Online Optimization with Delays: Asynchronicity, Adaptivity, and Optimism

Multi-Agent Online Optimization with Delays: Asynchronicity, Adaptivity, and Optimism

Online Distributed Coordinated Precoding for Virtualized MIMO Networks with Delayed CSI

Recent advances on distributed online optimization

Contact Info

Product

Resources

About