Heavy-ball: A new approach to tame delay and convergence in wireless network optimization

@article{Liu2016HeavyballAN,
  title={Heavy-ball: A new approach to tame delay and convergence in wireless network optimization},
  author={Jia Liu and Atilla Eryilmaz and Ness B. Shroff and Elizabeth Serena Bentley},
  journal={IEEE INFOCOM 2016 - The 35th Annual IEEE International Conference on Computer Communications},
  year={2016},
  pages={1-9}
}
  • Jia Liu, A. Eryilmaz, E. Bentley
  • Published 10 April 2016
  • Computer Science
  • IEEE INFOCOM 2016 - The 35th Annual IEEE International Conference on Computer Communications
The last decade has seen significant advances in optimization-based resource allocation and control approaches for wireless networks. [] Key Method Based on this heavy-ball technique, we develop a cross-layer optimization framework that offers utility-optimality, fast-convergence, and significant delay reduction. Our contributions are three-fold: i) we propose a heavy-ball joint congestion control and routing/scheduling framework for both single-hop and multi-hop wireless networks; ii) we show that the…

Figures from this paper

Achieving Low-Delay and Fast-Convergence in Stochastic Network Optimization: A Nesterovian Approach
  • Jia Liu
  • Computer Science
    SIGMETRICS 2016
  • 2016
TLDR
A new stochastic network optimization framework inspired by the Nesterov accelerated gradient method is developed that offers utility-optimality, fast-convergence, and significant delay reduction in stochastically network optimization.
Towards Fast-Convergence, Low-Delay and Low-Complexity Network Optimization
  • Sinong Wang, N. Shroff
  • Computer Science
    Abstracts of the 2018 ACM International Conference on Measurement and Modeling of Computer Systems
  • 2018
TLDR
A new algorithmic framework with fast convergence: it converges with only O(log(1/ε) iterations, that is the fastest speed among all the existing algorithms; and low delay: it guarantees optimal utility with finite queue length.
Towards Fast-Convergence, Low-Delay and Low-Complexity Network Optimization
TLDR
This paper builds on a kind of inexact Uzawa method in the Alternating Directional Method of Multiplier, and provides a new theoretical path to prove global and linear convergence rate of such a method without requiring the full rank assumption of the constraint matrix.
High-Order Momentum: Improving Latency and Convergence for Wireless Network Optimization
  • Jia Liu
  • Computer Science
    IEEE INFOCOM 2018 - IEEE Conference on Computer Communications
  • 2018
TLDR
This paper first proposes a new weight updating scheme that enables the incorporation of high-order momentum, and proves the throughput-optimality and queue-stability of the proposed high- order momentum-based approach and characterize its delay and convergence performances.
High-Order Momentum : Improving Latency and Convergence for Wireless Network Optimization
TLDR
This paper first proposes a new weight updating scheme that enables the incorporation of high-order momentum, and proves the throughput-optimality and queue-stability of the proposed highorder momentum-based approach and characterize its delay and convergence performances.
Momentum-Based Online Cost Minimization for Task Offloading in NOMA-Aided MEC Networks
TLDR
A momentum-based SGD method is proposed by amending SGD iterations with momentum terms, which can efficiently accelerate algorithm convergence while reducing the task queue backlog without loss of cost-optimality.
Scheduling algorithm design in multiuser wireless networks
  • Yi Chen
  • Business, Computer Science
  • 2016
TLDR
The Head-of-Line (HOL) access delay based scheduling algorithm (HAD) for flow-level dynamic systems is proposed, and it is shown that HAD is able to obtain throughput-optimality which is validated by simulation.
Learn-and-Adapt Stochastic Dual Gradients for Network Resource Allocation
TLDR
A novel learn-and-adapt stochastic dual gradient method is developed in this paper to learn the sample-optimal Lagrange multiplier from historical data, and accordingly adapt the upcoming resource allocation strategy.
Head-of-Line Access Delay-Based Scheduling Algorithm for Flow-Level Dynamics
TLDR
A head-of-line access delay (HAD)-based scheduling algorithm is proposed and it is shown that it is throughput-optimal when the flows are dynamic, i.e., they are short-lived with finite data to transmit.
A Theory of Second-Order Wireless Network Optimization and Its Application on AoI
TLDR
This paper introduces a new theoretical framework for optimizing second-order behaviors of wireless networks, and shows that this framework provides a very accurate characterization of AoI, and leads to a tractable scheduling policy that outperforms other existing work.
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 25 REFERENCES
Joint Congestion Control and Routing Optimization: An Efficient Second-Order Distributed Approach
TLDR
A new second-order joint congestion control and routing framework based on a primal-dual interior-point approach that offers utility-optimality, queue-stability, fast convergence, and low delay is proposed.
A tutorial on cross-layer optimization in wireless networks
TLDR
It is shown that a clean-slate optimization-based approach to the multihop resource allocation problem naturally results in a "loosely coupled" cross-layer solution, and how to use imperfect scheduling in the cross- layer framework is demonstrated.
Super-Fast Delay Tradeoffs for Utility Optimal Fair Scheduling in Wireless Networks
  • M. Neely
  • Computer Science
    Proceedings IEEE INFOCOM 2006. 25TH IEEE International Conference on Computer Communications
  • 2006
TLDR
This work demonstrates that the problem of maximizing throughput utility in a data network is fundamentally different than related problems of minimizing average power expenditure, as these latter problems cannot achieve such performance tradeoffs.
The impact of imperfect scheduling on cross-Layer congestion control in wireless networks
TLDR
This paper studies both the case when the number of users in the system is fixed and the case with dynamic arrivals and departures of the users, and establishes performance bounds of cross-layer congestion control with imperfect scheduling.
A Distributed CSMA Algorithm for Throughput and Utility Maximization in Wireless Networks
TLDR
An adaptive carrier sense multiple access (CSMA) scheduling algorithm that can achieve the maximal throughput distributively and is combined with congestion control to achieve the optimal utility and fairness of competing flows.
Distributed cross-layer optimization in wireless networks: A second-order approach
TLDR
This work makes the first known attempt to develop a distributed Newton's method, which is second-order and enjoys a quadratic convergence rate, and contributes to a potential second- order paradigm shift in wireless networks optimization theory.
Multi-Step Gradient Methods for Networked Optimization
TLDR
This work develops multi-step gradient methods for network-constrained optimization of strongly convex functions with Lipschitz-continuous gradients and applies the proposed technique to three engineering problems: resource allocation under network-wide budget constraint, distributed averaging, and Internet congestion control.
Joint congestion control, routing, and MAC for stability and fairness in wireless networks
TLDR
The main contribution is to prove the asymptotic optimality of a primal-dual congestion controller, which is known to model different versions of transmission control protocol well.
Q-CSMA: Queue-Length-Based CSMA/CA Algorithms for Achieving Maximum Throughput and Low Delay in Wireless Networks
TLDR
A discrete-time distributed randomized CSMA algorithm based on a generalization of the so-called Glauber dynamics from statistical physics, which generates collision-free transmission schedules while explicitly taking collisions into account during the control phase of the protocol, thus relaxing the perfect CSMA assumption.
Fairness and optimal stochastic control for heterogeneous networks
TLDR
The combined strategy is shown to yield data rates that are arbitrarily close to the optimal operating point achieved when all network controllers are coordinated and have perfect knowledge of future events.
...
1
2
3
...