Thomas Bauschert

Network planning under demand uncertainty with robust optimization

The planning of a communication network inevitably depends on the quality of both the planning tool and the demand forecast used. In this article, we show by example how the emerging area of robust optimization can advance the network planning by a more accurate mathematical description of the demand uncertainty. After a general introduction of the concept and its application to a basic network design problem, we present two applications: multi-layer and mixed-line-rate network design. We conclude with a discussion of extensions of the robustness concept to increase the accuracy of handling uncertainties.

Energy efficiency and CAPEX minimization for backbone network planning: Is there a tradeoff?

Energy efficiency and the minimization of capital expenditures (CAPEX) for network equipment are studied for two network architectures: IP over WDM and IP over OTN over WDM. Case study results from mixed integer linear programming (MILP) formulations indicate that a CAPEX optimized network yields solutions having minimum energy consumption for both architectures. However, it is shown that the most cost efficient architecture is not always the least energy consuming. We quantify the occurring tradeoff and observe that network equipment CAPEX are dominating in terms of cost. Additionally, it is observed that the relative power contribution of different network layers is independent of the average inter-node traffic demand for both studied architectures.

Mobile core network virtualization: A model for combined virtual core network function placement and topology optimization

This paper addresses an important aspect of mobile core network virtualization: The combined optimization of the virtual mobile core network topology (graph) and its embedding onto a physical substrate network. Basically this comprises the placement of mobile core virtual network functions (VNFs) onto the nodes of the physical substrate network, the determination of the interconnections towards the radio access network (RAN) and the Internet as well as the traffic routing between the VNFs. This problem differs from the traditional virtual network embedding (VNE) problem as the virtual network topology is not known in advance and several additional constraints apply, e.g. not every node of the physical substrate network might be able to host every VNF. We assume that the topology, link capacities and node resources of the physical substrate network are predefined and that a node comprises both packet forwarding and datacenter/server functionality. The node capabilities are defined by the processing, storage and switching (throughput) resources as well as the ability to host specific mobile core VNFs, i.e. the SGW, PGW, MME and HSS virtual functions. For the traffic routing, explicit single path routing is assumed. We propose a novel integer linear programming formulation which combines the optimization of the virtual network topology with VNE optimization. Optimization target is to minimize the cost of occupied link and node resources. Our formulation relies on the joint embedding of individual core network service chains where a core network service chain denotes the sequence of mobile core VNFs a user or control plane traffic flow traverses. We evaluate our model by means of two physical network topology examples taken from SNDlib [1]. It is shown that our approach outperforms traditional VNE optimization approaches in terms of optimality and computation time.

Carrier-grade Ethernet for packet core networks

Ethernet is a permanent success story, extending its reach from LAN and metro areas now also into core networks. 100 Gbit/s Ethernet will be the key enabler for a new generation of true end-to-end carrier grade Ethernet networks. This paper first focuses on functionality and standards required to enable carrier-grade Ethernet-based core networks and possible Ethernet backbone network architectures will be discussed. The second part then evaluates the CAPEX and OPEX performance of Ethernet core networks and competitive network architectures. The results propose that Ethernet will not only soon be mature enough for deployment in backbone networks but also provide huge cost advantages to providers. A novel complete, cost-effective and service-oriented infrastructure layer in the area of core networks will arise. The industry-wide efforts to cover remaining challenges also confirm this outlook.

Combined Virtual Mobile Core Network Function Placement and Topology Optimization with Latency Bounds

In this paper, a novel mathematical optimization model for virtual mobile core network embeddings with respect to latency bounds is presented. This formulation differs from the classical virtual network embedding (VNE) model as the virtual network topology (i.e. The allocation of e.g. eNodeBs to single core gateways) is not known in advance but subject to optimization. Our formulation can be regarded as network functions/service chaining approach [1] as it relies on the joint embedding of individual core network service chains where a core network service chain denotes the sequence of mobile core virtual network functions (VNFs) a user or control plane traffic flow traverses. Regarding the placement decision of virtual core network functions, we consider upper bounds for the latency caused by processing, packet queueing and propagation. It is assumed that the queueing and processing delay depends on the user/control plane traffic utilization of the node/virtual machine on which the respective VNF is executed and that the propagation delay is proportional to the path length of the respective user/control plane traffic flow. The performance of the proposed optimization model is evaluated for the case of the European example network topology NOBEL-EU as physical substrate network, taken from SNDlib [2].

Quantifying CAPEX savings of homing architectures enabled by future optical network equipment

Alternative homing architectures can lead to significant cost reductions. These reductions are quantified using linear programming under a multi-layer consideration. In this study we focus on savings with respect to network equipment capital expenditures (CAPEX) achieved by transitioning to different homing architectures in an IP over optical data unit (ODU) setting. The impact of traffic demand is examined for different reference networks. In all conducted case studies we find that network equipment costs follow an approximately linear relationship with the traffic demand. Additionally, case study results assuming up to an aggressive 50% cost decrease in IP layer equipment, quantify the impact of variations in the cost of network elements.

Combining multi-period and multi-layer network planning: Ignored potential?

Multi-layer and multi-period network planning can lead to significant cost reductions. Using linear programming we study the effect of their joint consideration for IP-over-WDM networks. Case studies show cost savings surpassing 20% by including forecast knowledge in multi-layer optimization.

CAPEX and availability tradeoffs of homing architectures in multi-layer networks

A generic multi-layer model and a linear programming (LP) formulation are developed enabling the calculation of the optimal solution in terms of network equipment capital expenditures (CAPEX) in multi-homing design. Alternative homing architectures to traditional dual homing are studied not only with respect to network equipment costs but also regarding the achieved availability gain. Case study results show that significant CAPEX savings can be obtained by deploying a shared router strategy, which are independent from the traffic demand and deployed equipment.

Shared backup router resources: realizing virtualized network resilience

The emergence of multilayer networking capabilities opens the path for the deployment of advanced network architectures and resilience concepts. In this article we propose a novel resource-efficient homing scheme we call dual homing with shared backup router resources (SBRR). As the scheme realizes shared routerlevel redundancy with the help of dynamic circuit-switched services, it can achieve significantly lower network costs than todays dual homing architectures. We focus on the discussion of operational aspects related to control plane functions and the evaluation of cost savings. In addition, we address further issues such as the impact of the new homing architecture on the end-to-end connection availability and energy efficiency.

On the dimensioning of voice over IP networks for various call admission control schemes

In this paper we investigate various call admission control (CAC) schemes for Voice over IP (VoIP) architectures in the context of network dimensioning. For each of the considered CAC concepts we introduce an abstract model and propose an optimized dimensioning procedure. Based on numerical results for several network scenarios, the trade-off between IP network overdimensioning on one side and the complexity of the CAC scheme on the other side is demonstrated.