Catálogo de publicaciones - libros

Next generation mobile communication systems will comprise both WLAN technologies and ad hoc networks. Ad hoc networks are formed by the spontaneous collaboration of wireless nodes. When communication to the Internet is desired, one or more nodes must act as gateways for the ad hoc network. In this case, global addressing of ad hoc nodes is required. In this paper, we present an IPv6 addressing architecture in order to be able to support “pure” spontaneous IPv6 ad hoc networks but also to allow seamless integration between wireless LANs and ad hoc networks. It implies the possibility to discover a gateway/prefix pair which is used in order to build an IPv6 global address and, when necessary, to maintain a default route towards the Internet.

We present the framework of SDSR, a Scalable, efficient, robust and load-balanced Data Storage/Retrieval service for large scale wireless ad hoc networks. SDSR hashes each data key to normalized geographical coordinates (x,y), which can be seen as a point for storing/retrieving the data in a unit grid. SDSR achieves scalability, robustness and load-balancing by partitioning the network into hierarchical grids of increasing sizes and replicating the data item into each grid. The storage location in each grid is determined by scaling the normalized coordinates to the corresponding grid size. SDSR retrieves a data item in the same way as it stores the data item. We show that in query dominant, large scale wireless ad hoc networks, SDSR performs better than existing schemes in terms of energy efficiency, query latency, hotspot usage, and resilience to clustering failures. It scales well when the network size and the number of queries increase.

In order for multicast data packets to be transmitted efficiently, the multicast data forwarding is required at each router. Within a wired network, a network interface of a node has a one-to-one connection to one of the incoming interfaces of other nodes. If the packet coming into an incoming interface needs to be forwarded, it is delivered to the corresponding outgoing interface. However, in a mobile ad hoc network (MANET), in most cases each node has only one network interface, so the incoming interface is the same as the outgoing one and the wireless network interface of a node has a one-to-many connection to those of neighboring nodes. This difference may cause problems such as routing loops and packet duplication. Therefore, in this paper, we propose a multicast data forwarding scheme which can be used in the multi-hop wireless ad hoc network without causing either routing loops or packet duplication. In the proposed scheme, a table is defined for the prevention of packet duplication that can happen when the tree-based multicast routing protocol is used. We have implemented our proposed scheme by using the netfilter [1] of the Linux OS.

We propose a global design procedure for heterogeneous networks , which includes the definition of their topology; routing procedures; link capacity assignment; transfer mode; and traffic policy. We discuss the network model, which minimizes the cost of interconnecting a number of nodes whose locations are known; the traffic model, where we use the concept of Equivalent Bandwidth, and the resolution algorithms, where we compare a Simulated annealing algorithm (SAA) with a commercial solver, with good results. Results show that SAA gets to the optimum solution over 10 times faster than the commercial solver. Experiments also show that, for networks with more than 50 nodes, the SAA still delivers good feasible solutions while the commercial solver is unable to deliver results.

Ternary Content-Addressable Memories (TCAMs) provide a fast mechanism for IP lookups and a simple management for route updates. The high power consumption problem can be resolved by providing a TCAM partitioning technique that selectively addresses smaller portions of a TCAM. This paper proposes a 2-level TCAM architecture using prefix comparison rule to partition the routing table into a number of buckets of possibly equal sizes. The 2-level architecture results in a 2-step lookup process. The first step determines which bucket is used to search the input IP. The second step only searches the IP in the determined bucket from the first step. The prefix partitioning algorithm provides an incremental update. Experiments show that the proposed algorithm has lower power consumption than the existing TCAM partitioning algorithms.

In this paper, we study the hardness of the IP-subnet aware routing problem in WDM network. IP-subnet aware routing attempts to reduce routing overhead by grouping a set of addresses under a single subnet. However, routing in WDM network with subnets imposes neighboring IP interfaces along the path to be on the same subnet. This subnet constraint forces the routing path to be longer unless routing permits reconfiguration of subnets. The hardness of this routing problem is first studied in [1]. We investigate further the problem of finding paths accounting for subnets and its hardness. Regardless of subnets, we argue that a shortest IP hop path can be constructed in a polynomial time. We provide efficient routing algorithms to justify our argument.

Embedded real-time systems are deployed in a wide range of application domains such as transportation systems, automated manufacturing, home appliances and telecommunications. Originally developed for use in automotive applications, CAN (Controller Area Network) has been adopted by industry as the standard network technology to transmit data for sensors and actuators, due to its fast response and high reliability for applications. We design and implement a logical communication model and a real-time data transmission protocol for embedded systems with CAN. This logical communication model with prioritized data transmission is effective for developing application programs in an embedded system. Owing to the small size of the CAN frame format, there are many restrictions on the implementation of a network protocol on CAN. In an attempt to solve this problem, we describe a concrete implementation of the network protocols in detail to realize the logical communication model.

In this paper, we propose DINPeer middleware to overcome limitations in current peer-to-peer (P2P) overlay systems. DINPeer exploits a spiral-ring method to discover an inner ring with relative largest bandwidth to form a DINloop (Data-In-Network loop). DINPeer further integrates DINloop with P2P overlay network via node state and routing algorithm. The key features of DINPeer include using DINloop to replace a multicast rendezvous point and turning DINloop into a cache to achieve data persistency. Simulations show that DINPeer is able to optimize P2P communication in a number of ways, such as when the size of the DINloop is capped within a limit, it can achieve a better performance than native IP multicast and P2P overlay multicast systems.

To minimize the network resource and to meet the need of related application programs such as Internet Broadcasting or Video conferencing are the main concerns of Host-based Multicast tree construction algorithm. Existing works reduced performance-degrading factors like the duplicate data transmission on the same network link and overhead incurred at host or end-systems. However, they had relatively high RDP(Relative Delay Penalty), which made it difficult to adapt for application programs. In this work, we proposed a DDTA(Data Delivery Tree Adjust) algorithm which can reduce RDP by minimizing the tree depth, but it also could incur performance debasement. Proposed algorithm adapted techniques such as the detection of hosts existing in the same LAN to decrease the RDP and node switching to compensate performance loss. In addition, it suggests schemes for rapid recovery of a data delivery tree and shows that its RDP is lower than existing work.

To reduce perceived impact, we argue that, in P2P streaming networks, selection of new parent peer for a peer should consider not only network quality (e.g. delay and bandwidth) but also the frame-buffer status between the parent-child peers. When there is large mismatch on the frame-buffer, the child peer is in danger of suffering frame-buffer underflow while playback. Meanwhile, another issue discussed in this paper is phase-skew among peers. Degrees of phase-skew between two peers mean the degrees of presentation-time difference between the two peers. Without proper control, phase-skew could be serious between peers. In this paper we propose an effective buffer management and coordination scheme to avoid these two problems. A prototype P2P streaming system basing on open standards was built to verify our design approaches.