Yangming Zhao

Load balance vs energy efficiency in traffic engineering: A game Theoretical Perspective

In this paper, we study the tradeoff between two important traffic engineering objectives: load balance and energy efficiency. Although traditional commonly used multi-objective optimization methods can yield a Pareto efficient solution, they need to construct an aggregate objective function (AOF) or model one of the two objectives as a constraint in the optimization problem formulation. As a result, it is difficult to achieve a fair tradeoff between these two objectives. Accordingly, we induce a Nash bargaining framework which treats the two objectives as two virtual players in a game theoretic model, who negotiate how traffic should be routed in order to optimize both objectives. During the negotiation, each of them announces its performance threat value to reduce its cost, so the model is regarded as a threat value game. Our analysis shows that no agreement can be achieved if each player sets its threat value selfishly. To avoid such a negotiation break-down, we modify the threat value game to have a repeated process and design a mechanism to not only guarantee an agreement, but also generate a fair solution. In addition, the insights from this work are also useful for achieving a fair tradeoff in other multi-objective optimization problems.

A New Heuristic for Monitoring Trail Allocation in All-Optical WDM Networks

We study the m-trail (monitoring trail) allocation problem in all-optical WDM mesh networks for achieving fast and unambiguous link failure localization. The existing ILP is not feasible for solving the problem in large-size networks. A heuristic RCA+RCS can find feasible solutions in a shorter running time, but it is a randomized algorithm. More importantly, RCA+RCS suffers from the disjoint trail problem which dramatically increases the number of required monitors in large-size networks. In this paper, we propose a new heuristic MTA (Monitoring Trail Allocation) to solve the problem. MTA avoids those issues in RCA+RCS, and achieves an efficient tradeoff between monitor cost and bandwidth cost. Compared with RCA+RCS, MTA greatly shortens the running time and achieves a much higher solution quality. We also show that MTA provides a flexible framework to enable multiple possible variations for future study.

Minimizing average coflow completion time with decentralized scheduling

In current data centers, an application (e.g. MapReduce) usually generates a collection of parallel flows sharing a common goal. These flows compose a coflow and only completing them all is meaningful. Accordingly, minimizing the average coflow completion time (CCT) becomes a critical objective for flow scheduling. In this topic, the state-of-the-art centralized method, Varys, achieves a good average CCT; but it has the scalability problem. Alternatively, the only existing decentralized method, Baraat, suffers from the head-of-line blocking problem. To solve these problems, we propose D-CAS, a preemptive, decentralized, coflow-aware scheduling system in this paper. D-CAS pursues coflow-level remaining-time-first (MRTF) principle by leveraging a simple negotiation mechanism between each coflows data senders and receivers. As the MRTF principle is inherently preemptive and proven to be a near-optimal guideline to minimize average CCT, D-CAS avoids the head-of-line blocking problem and gets good performances. Through extensive simulations, we find that D-CAS achieves a performance close to Varys (gap <; 15%) and outperforms Baraat significantly (about 1.4-4×).

Virtual network mapping for multicast services with max-min fairness of reliability

Network function virtualization (NFV) provides an effective way to reduce the network providers cost by allowing multiple virtual networks (VNs) to share the underlying physical infrastructure. In the NFV environment, especially when supporting multicast services over the VNs, reliability is a critical requirement since the failure of one virtual node can cause the malfunction of multiple nodes that receive multicasting data from it. In this paper, we study for the first time to the best of our knowledge how to efficiently map VNs for multicast services over both general IP networks and orthogonal frequency division multiplexing (OFDM)-based elastic optical networks (EONs) while taking into consideration the max-min fairness in terms of reliability among distinct VNs. For general IP networks, we propose a mixed integer linear programming (MILP) model to determine the upper bound on the reliability with max-min fairness. In addition, an efficient heuristic, namely a reliability-aware genetic (RAG) algorithm, is developed to address reliable multicast VN mapping with a low computational complexity. By encoding multicast tree construction and link mapping into the process of path selection, taking into consideration the reliability with max-min fairness, and the networking reliability factors during mutation, RAG can globally optimize the reliability and fairness of all the multicast VN requests. For OFDM-based EONs, we extend the MILP (RAG) to optical-MILP [(O-MILP) optical RAG (O-RAG)] by considering the most efficient modulation format selection strategy, spectrum continuity, and conflict constraints. Through extensive simulations, we demonstrate that RAG (O-RAG) achieves close to the optimal reliability fairness with a much lower time complexity than the MILP (O-MILP) model. In particular, the path reliability-based mutation strategy in RAG (O-RAG) yields a significant performance improvement over other heuristic solutions in terms of reliability fairness, bandwidth (spectrum) consumption, and transmission delay.

Towards Practical and Near-Optimal Coflow Scheduling for Data Center Networks

In current data centers, an application (e.g., MapReduce, Dryad, search platform, etc.) usually generates a group of parallel flows to complete a job. These flows compose a coflow and only completing them all is meaningful to the application. Accordingly, minimizing the average Coflow Completion Time (CCT) becomes a critical objective of flow scheduling. However, achieving this goal in todays Data Center Networks (DCNs) is quite challenging, not only because the schedule problem is theoretically NP-hard, but also because it is tough to perform practical flow scheduling in large-scale DCNs. In this paper, we find that minimizing the average CCT of a set of coflows is equivalent to the well-known problem of minimizing the sum of completion times in a concurrent open shop. As there are abundant existing solutions for concurrent open shop, we open up a variety of techniques for coflow scheduling. Inspired by the best known result, we derive a 2-approximation algorithm for coflow scheduling, and further develop a decentralized coflow scheduling system, D-CAS, which avoids the system problems associated with current centralized proposals while addressing the performance challenges of decentralized suggestions. Trace-driven simulations indicate that D-CAS achieves a performance close to Varys, the state-of-the-art centralized method, and outperforms Baraat, the only existing decentralized method, significantly.

Multiobjective Optimization for Green Network Routing in Game Theoretical Perspective

In this paper, we study the multiobjective optimization problem for green network routing. Although traditional commonly used multiobjective optimization methods can yield a Pareto efficient solution, they need to construct an aggregate objective function (AOF) or model one objective as a constraint in the optimization problem formulation. As a result, it is difficult to achieve a fair tradeoff among all objectives. Accordingly, we induce a Nash bargaining framework, which treats the two objectives as two virtual players in a game theoretic model, who negotiate how traffic should be routed to optimize both objectives. During the negotiation, each of them announces its performance threat value to reduce its cost, so the model is regarded as a threat value game. Our analysis shows that no agreement can be achieved if each player sets its threat value selfishly. To avoid such a negotiation break-down, we modify the threat value game to have a repeated process and design a mechanism to not only guarantee an agreement, but also generate a fair solution. Finally, to evaluate the efficiency of our proposed framework, we implement it into two multiobjective optimization cases for network green routing. The first case is load balancing and energy efficiency optimization for intradomain routing, and the second one is the energy efficiency optimization of two domains for interdomain routing.

Towards optimal outsourcing of service function chain across multiple clouds

As Network Function Virtualization (NFV) becomes reality and cloud computing offers a scalable pay-as-you-go charging model, more network operators would like to outsource their Service Function Chains (SFC) to the public clouds in order to reduce the operational cost. However, how to minimize the operational cost with Quality of Service (QoS) guarantee when outsourcing SFC is still an open problem. In this paper, we are to study this problem when there are large number of candidate cloud providers with diverse pricing schemes of network functions. In addition, extra delay is introduced as the result of outsourcing SFCs. Firstly, we formulate this problem as an Integer Linear Programming (ILP) model. Then we design an efficient heuristic algorithm named QoS-Guaranteed SFC Outsourcing algorithm (QGSO) based on Hidden Markov Model (HMM). The extensive simulations show that QGSO saves up to 75.8% cost compared with that of deploying network functions in local network. QGSO also achieves up to 42.6% cost savings compared with the result of first-fit based optimization algorithm.

Virtual network mapping for reliable multicast services with max-min fairness

Network Function Virtualization (NFV) provides an effective way to reduce the network providers cost by allowing multiple Virtual Networks (VNs) to share the underlying physical infrastructure. In the NFV environment, especially when supporting multicast service over the VNs, reliability is a critical requirement in the process of VN mapping since the failure of one virtual node can cause the malfunction of all the subsequent nodes that receive multicasting data from it. In this paper, for the first time, we study how to efficiently map VNs for reliable multicast services, while taking into consideration the max-min fairness of the reliability among distinct VNs. We propose a Mixed Integer Linear Programming (MILP) model to determine the upper bound on the max-min fairness reliability. In addition, an efficient heuristic, namely Uniform Reliability Mutation based Genetic (URMG) algorithm, is developed to address reliable multicast VN mapping with a low computational complexity. By encoding multicast tree construction and link mapping into path selection, taking into consideration the max-min reliability fairness goal, and the networking reliability factors during mutation, URMG can globally optimize the reliability and its fairness of all the multicast VN requests. Through extensive simulations, we demonstrate that URMG achieves close to the optimal reliability fairness with a much lower time complexity than the MILP and yields a significant performance improvement in terms of reliability fairness, bandwidth consumption and transmission delay comparing with other heuristic solutions.

Monitoring Trail Allocation in all-optical networks with the Random Next Hop Policy

The concept of monitoring trail (m-trail) provides a striking mechanism for fast and unambiguous link failure localization in all-optical networks. To achieve fast m-trail design in large-size networks, two efficient heuristics RCA+RCS and MTA are proposed against the optimal ILP (Integer Linear Program) model. However, RCA+RCS suffers from the disjoint trail problem which increases the required number of m-trails, and MTA always finds a deterministic solution which may not be good enough due to the limited solution space. In this paper, we propose a new heuristic RNH-MTA (Monitoring Trail Allocation with the Random Next Hop policy) to solve those issues. Similar to MTA, RNH-MTA ensures a valid optical structure of each m-trail and sequentially adds necessary m-trails to the solution, and thus is free of the disjoint trail problem. By replacing the deterministic searching in MTA using the Random Next Hop policy, RNH-MTA sets up a probabilistic model in extending each m-trail. This not only enlarges the solution space and increases the solution diversity, but also enables a controllable tradeoff between the solution quality and the running time of the algorithm. Our numerical results show the advantages of RNH-MTA over both RCA+RCS and MTA.

On Progressive Recovery in Interdependent Cyber Physical Systems

This paper studies how to determine an optimal order of recovering interdependent Cyber Physical Systems (CPS) after a large scale failure. In such a CPS, some failed devices must be repaired first before others can. In addition, such failed devices require a certain amount of repair resources and may take multiple stages to repair. We consider two scenarios: 1) reserved model where all the required repair resources should be prepared at the beginning of repairing a device; and 2) opportunistic model where we can partially repair a device with only part of the required resources. For each scenario, we model it using an Integer Linear Programming (ILP) and use a relaxation and rounding method to design an ILP based algorithm. In addition, we also design a Dynamic Programming (DP) based algorithm. Simulation results show that ILP based algorithm outperforms DP based algorithm by 10%-20% in systems with less than 200 failed devices, but DP based algorithm can support extreme large size systems with more than 5000 failed devices.