Abstract — Ad hoc networks are collections of mobile nodes communicating using wireless media, without any fixed infrastructure. Existing multicast protocols fall short in a harsh ad hoc mobile environment because node mobility causes conventional multicast trees to rapidly become outdated. The amount of bandwidth resources required for building up a multicast tree is commonly less than that required for other delivery structures, since a tree avoids unnecessary duplication of data. However, a tree structure is more subject to disruption due to link/node failure and node mobility than more meshed structures. This paper explores these contrasting issues, and proposes PPMA, a new probabilistic predictive multicast algorithm in ad hoc networks, which leverages the tree delivery structure for multicasting, solving its drawbacks in terms of lack of robustness and reliability in highly mobile environment. PPMA overcomes the existing tradeoff between the bandwidth efficiency to set up a multicast tree, and the tree robustness to node energy consumption and mobility, by decoupling tree efficiency from mobility robustness. By exploiting the non-deterministic nature of ad hoc networks, the proposed algorithm takes into account the estimated network state evolution in terms of node residual energy, link availability, displacement of set up multicast trees, and node mobility forecast, in order to maximize the multicast tree lifetime. The algorithm statistically tracks the relative movements among nodes to capture the dynamics in the ad hoc network. This way, PPMA estimates the nodes’ future relative positions, in order to calculate a long lasting multicast tree. To do so, it exploits those more stable links in the network, while minimizing the total network energy consumption. We propose PPMA in both its centralized and distributed version, providing performance evaluation through extensive simulation campaign run on an ad hoc C++ based simulator. I.

Multiprotocol Label Switching (MPLS) can improve the routing efficiency of Internet Protocol (IP) networks through its intrinsic Traffic Engineering (TE) capabilities. In this paper, a centralised and a distributed virtual-flow routing algorithms are proposed, which aggregate IP flows entering the MPLS domain and optimally partition them among virtual flows that are routed on multiple paths. The routing algorithms dynamically select multiple Label Switched Paths (LSPs), taking into account the available bandwidth of links in the network to balance the traffic load and avoid network congestion. The multipath routing problem is formulated as a Multicommodity Network Flow (MCNF) problem, and is solved by implementing online the Dantzig–Wolfe decomposition method. The proposed multipath algorithms are shown through simulations to outperform single-path routing solutions, such as the Constraint Shortest Path First (CSPF) and the Bandwidth-Based Shortest Path (BSPR) routing algorithms.