IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 58, NO. 8, OCTOBER 2009 4447 Multicast QoS Core-Based Tree Routing Protocol and Genetic Algorithm Over an HAP-Satellite Architecture Floriano De Rango, Member, IEEE, Mauro Tropea, Amilcare Francesco Santamaria, and Salvatore Marano, Member, IEEE Abstract—In this paper, a quality-of-service (QoS) multicast routing protocol, i.e., the core-based tree based on heuristic ge- netic algorithms (GAs), has been implemented and applied over a high-altitude platform (HAP)-satellite platform. The proposed multicast routing algorithm is called the constrained cost- bandwidth-delay GA (CCBD-GA). To achieve a better optimiza- tion of the multicast tree cost, a new algorithm called HULK-GA, which is based on the GA and on a proposed broadcast metric, has been developed. Finally, an algorithm called hybrid cost- bandwidth-delay GA has been proposed, taking into account both CCDB-GA and HULK-GA characteristics to obtain an overall algorithm that can consider QoS routing constraints and minimize the overall cost per link of the considered multicast tree. The joint bandwidth-delay metrics can be very useful in hybrid platforms such as the platform considered, because it is possible to take advantage of the single characteristics of the satellite and HAP seg- ments. The HAP segment offers low propagation delay, permitting QoS constraints based on maximum end-to-end delay to be met. The satellite segment, instead, offers a larger footprint but higher propagation delay. The joint bandwidth-delay metric permits the traffic load to be balanced, respecting both QoS constraints. Index Terms—Bandwidth, delay, digital video broadcasting- return channel via satellite (DVB-RCS), genetic algorithms (GAs), high-altitude platform (HAP), multicast routing, quality of service (QoS). I. I NTRODUCTION A NUMBER of emerging network applications require the delivery of packets from one or more senders toward a group of receivers. These applications include data transfer (e.g., the transfer of software upgrades from the software houses to end users), continuous streaming media (e.g., mul- timedia contents), shared data applications (e.g., a whiteboard or teleconferencing applications that are shared among many distributed participants), interactive gaming, and so on. Multi- cast manages a tree to distribute the data flow into the network. In this structure, the routers represent the multicast tree nodes Manuscript received August 8, 2008; revised February 6, 2009. First pub- lished March 24, 2009; current version published October 2, 2009. The review of this paper was coordinated by Prof. M. Guizani. F. De Rango and S. Marano are with the Department of Electronics, Infor- matics, and Systems, University of Calabria, 87036 Arcavacata di Rende, Italy (e-mail: derango@deis.unical.it; marano@deis.unical.it). M. Tropea and A. F. Santamaria are with the Telecommunications Research Group, Department of Electronics, Informatics, and Systems, University of Calabria, 87036 Arcavacata di Rende, Italy (e-mail: mtropea@deis.unical.it; afsantamaria@deis.unical.it). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TVT.2009.2019281 that permit connection with sources with destinations. It is clear that the use of IP Multicast is a serious gain when the same data has to be sent to a number of receivers. The problem with IP multicast is that not all IP networks have enabled IP Multicast, and this condition means that the majority of end users cannot receive the IP Multicast services that require mainly the mul- ticast paradigm, e.g., audio and video on demand and shared file contents. Multicast routing protocols and algorithms are designed to construct and maintain a tree structure; moreover, they provide for complete support to member joining/leaving and maintaining updated information. Moreover, due to QoS issues, many multicast protocols have been considered in the literature [1]–[6]. QoS IP multicast involves the construction and maintenance of trees that reflect a desired service quality. Most deployed interdomain and intradomain unicast routing protocols use a single metric based on the shortest number of hops between two nodes to calculate routes. Multicast proto- cols are associated with multicast algorithms, with the main task of building the multicast tree and, moreover, updating or changing it, if necessary. There are several kinds of algorithms in the literature. The classical approach is based on polynomial algorithms such as Dijkstra or Prim, which resolve problems when only a QoS metric is considered [3], [7], [8]. When a more complex problem must be faced, e.g., the Steiner tree problem (STP), other types of algorithms, which can solve these multicast problems, must be considered. These approaches can satisfy multiconstraint end-user requests. The genetic algorithm (GA) is a specific metaheuristic that can address the multicast multiconstraint problems [5], [9]–[17], permitting a suboptimal solution to be found. In this paper, a multicast routing algorithm based on the GA is proposed, called the constrained cost- bandwidth-delay GA (CCBD-GA), which provides a method for solving the QoS constrained multicast routing problem [4]. The proposed GA-based multicast routing algorithm per- formances were evaluated through simulations under different network topologies. Moreover, to take advantage of the consid- ered network topology, another cost index was evaluated and deployed. In particular, the broadcast nature of the network permits the cost of every link on several users to be amortized. This cost is called broadcast cost, and this idea is called the broadcast gain (BG) concept. This type of algorithm is intro- duced in an extended core-based tree (E-CBT) protocol [3], [4], [6]. The E-CBT is a QoS-driven protocol, and it can consider additive and concave metrics. Moreover, the test platform was chosen to be a hybrid platform with one on-board processor 0018-9545/$26.00 © 2009 IEEE