Network virtualization and slicing technologies create opportunity for infrastructure-less virtual network operators (VNOs) to enter the market anytime and provide diverse services. Multiple VNOs compete to provide the same kinds of services to end users (EUs). VNOs lease virtual resources from the infrastructure provider (InP) and sell services to the EUs by using the leased resources. The difference between the selling and leasing is the gross profit for the VNOs. A VNO that leases resources without precise knowledge of future demand, may not consume all the leased resources through service offers to EUs. Consequently, the VNO experiences loss and resources remain unused. In order to improve resource utilization and ensure that new entrant VNOs do not face losses, proper estimation of initial resource demand is important. In this paper, we propose a Bayesian game with Cournot oligopoly model to properly estimate the optimal initial resource demands for multiple entrant competing VNOs (players) with the objective of maximizing the expected profit for each VNO. The VNOs offer the same kinds of services to EUs with different qualities (player's type), which are public information. The exact service quality with which a VNO competes in the market is private information. Therefore, a VNO assumes the type of its opponent VNOs with certain probability. We derive the Bayesian Nash equilibrium (BNE) of the presented game and evaluate numerically the effect of service qualities and prices on the expected profit and market share of the VNOs.

1+1 protection provides instantaneous proactive recovery from any single link failure by duplicating and sending the same source data onto two disjoint paths. Other resource efficient recovery techniques to deal with single link failure require switching operations at least at both ends, which restrict instantaneous recovery. However, the 1+1 protection technique demands at least double network resources. Our goal is to minimize the resources required for 1+1 protection while maintaining the advantage of instantaneous recovery. It was reported that the network coding (NC) technique reduces resource utilization in 1+1 protection, and in order to determine an optimum NC aware set of routes that minimizes the required network resources for 1+1 protection, an Integer Quadratic Programming (IQP) formulation has already been addressed. Solving an IQP problem requires large amount of memory (cannot be determined exactly) and special algorithms by the mathematical programming solver. In this paper our contributions consist of two parts. First, we formulate the optimization problem, corresponding to the IQP model, as an Integer Linear Programming (ILP) formulation, which is solvable by any linear programming solver, and so its memory and time requirements are smaller. However, the presented ILP model works well in small-scale and medium-scale networks, but fails to support large-scale networks due to excessive memory requirements and calculation time. Second, to deal with these issues, a heuristic algorithm is proposed to determine the best possible NC aware set of routes in large-scale networks. Numerical results show that our strategies achieve almost double the resource saving effect than the conventional minimal-cost routing policy in the examined medium-scale and large scale networks.

Smart OSPF (S-OSPF), a load balancing, shortest-path-based routing scheme, was introduced to improve the routing performances of networks running on OSPF assuming that exact traffic demands are known. S-OSPF distributes traffic from a source node to neighbor nodes, and after reaching the neighbor nodes, traffic is routed according to the OSPF protocol. However, in practice, exact traffic demands are difficult to obtain, and the distribution of unequal traffic to multiple neighbor nodes requires complex functionalities at the source. This paper investigates non-split S-OSPF with the hose model, in which only the total amount of traffic that each node injects into the network and the total amount of traffic each node receives from the network are known, for the first time, with the goal of minimizing the network congestion ratio (maximum link utilization over all links). In non-split S-OSPF, traffic from a source node to a destination node is not split over multiple routes, in other words, it goes via only one neighbor node to the destination node. The routing decision with the hose model is formulated as an integer linear programming (ILP) problem. Since the ILP problem is difficult to solve in a practical time, this paper proposes a heuristic algorithm. In the routing decision process, the proposed algorithm gives the highest priority to the node pair that has the highest product of the total amount of injected traffic by one node and total amount of received traffic by the other node in the pair, where both traffic volumes are specified in the hose model, and enables a source node to select the neighbor node that minimizes network congestion ratio for the worst case traffic condition specified by the hose model. The non-split S-OSPF scheme's network congestion ratios are compared with those of the split S-OSPF and classical shortest path routing (SPR) schemes. Numerical results show that the non-split S-OSPF scheme offers lower network congestion ratios than the classical SPR scheme, and achieves network congestion ratios comparable to the split S-OSPF scheme for larger networks. To validate the non-split S-OSPF scheme, using a testbed network experimentally, we develop prototypes of the non-split S-OSPF path computation server and the non-split S-OSPF router. The functionalities of these prototypes are demonstrated in a non-split S-OSPF network.

This paper proposes route advertisement policies (RAP) and an inbound traffic engineering (ITE) technique for a multihomed autonomous system (AS) employing the Border Gateway Protocol (BGP) and provider aggregatable (PA) addressing. The proposed RAP avail the advantage of address aggregation benefit of PA addressing. If multiple address spaces are allocated to each of the ASes that are multihomed to multiple upstream ASes, reduction of the forwarding information base (FIB) and quick convergence are achieved. However, multihoming based on PA addressing raises two issues. First, more specific address information is hidden due to address aggregation. Second, multiple allocated address spaces per AS does not provide the capability of ITE. To cope with these two limitations, we propose i) RAP to ensure connectivity among ASes with fewer routes installed in the FIB of each top-tier AS, and ii) an ITE technique to control inbound routes into multihomed ASes. Our ITE technique does not increase the RIB and FIB sizes in the Internet core. We implement the proposed RAP in an emulation environment with BGP using the Quagga software suite and our developed Hierarchical Automatic Number Allocation (HANA) protocols. We use HANA as a tool to automatically allocate hierarchical PA addresses to ASes. We confirm that with our proposed policies the FIB and RIB (routing information base) sizes in tier-1 ASes do not change with the increase of tier-3 ASes, and the number of BGP update messages exchanged is reduced by up to 69.9% from that achieved with conventional BGP RAP. We also confirmed that our proposed ITE technique, based on selective prefix advertisement, can indeed control inbound traffic into a multihomed AS employing PA addressing.

In this paper, we propose a method for automatic virtual resource allocation by using a multi-target classification-based scheme (MTCAS). In our method, an Infrastructure Provider (InP) bundles its CPU, memory, storage, and bandwidth resources as Network Elements (NEs) and categorizes them into several types in accordance to their function, capabilities, location, energy consumption, price, etc. MTCAS is used by the InP to optimally allocate a set of NEs to a Virtual Network Operator (VNO). Such NEs will be subject to some constraints, such as the avoidance of resource over-allocation and the satisfaction of multiple Quality of Service (QoS) metrics. In order to achieve a comparable or higher prediction accuracy by using less training time than the available ensemble-based multi-target classification (MTC) algorithms, we propose a majority-voting based ensemble algorithm (MVEN) for MTCAS. We numerically evaluate the performance of MTCAS by using the MVEN and available MTC algorithms with synthetic training datasets. The results indicate that the MVEN algorithm requires 70% less training time but achieves the same accuracy as the related ensemble based MTC algorithms. The results also demonstrate that increasing the amount of training data increases the efficacy ofMTCAS, thus reducing CPU and memory allocation by about 33% and 51%, respectively.