los servicios móviles de biblioteca
5. Mobiliario y equipamiento
5.1 Estándares .1 Estanterías
5.1.2 Mesa de trabajo
3.7
Summary and Conclusion
Vehicular ad hoc networks (VANETs) enable multi-hop vehicle-to-vehicle and vehicle-to-roadside wireless communication in a self-organized ad hoc network. The movement of vehicles represents the main characteristic of VANETs. Vehicu- lar mobility results in frequent topology changes. Novel routing protocols, such as position-based routing (PBR), are tailored to this environment since PBR forwards packets per hop, and no end-to-end route is required. Consecutive packets might follow different paths. These unique characteristics affect end-to-end connections, in particular transport protocols. This chapter analyzes the path characteristics that transport protocols experience in highway scenarios, such as connectivity and dis- ruption duration, packet loss, reordering, round trip time (RTT) and RTT jitter. These results aid in the following design of a vehicular transport protocol (VTP) which is evaluated through simulations.
The evaluation of the path characteristics investigates the metrics connectivity and disruption duration, packet loss probability and characteristics, packet reorder- ing, RTT and RTT jitter.
The connectivity evaluation results show that steady communication is feasible for source-destination distances up to 2000 m. For a distance of 2000 m, about 40% of the connections remain uninterrupted for 10 s on the average. With decreasing distance, the connectivity duration even increases. Disruptions resume after 3 s at the latest, only marginally dependent on the distance.
The packet loss ratio for a constant packet stream is huge: For a distance of 2000 m, standard PBR shows a packet loss rate of almost two thirds which can be significantly reduced to 22% when using cross-layer integration.
Although the RTT and RTT jitter are acceptably small for source-destination distances below 700 m, higher distances result in extreme fluctuation in RTT, e.g., up to 300% for a 2000 m distance.
Finally, reordering ratios for light loads are small (below 1%), but increase to 15% for medium data loads.
These unique characteristics of VANETs necessitate the development of a novel transport protocol.
The evaluation results of the path characteristics influence the design of a vehic- ular transport protocol (VTP) that is tailored to the unique properties of VANETs. VTP aims at maximizing the throughput of a connection while preserving fairness to competing data traffic. The objectives of VTP include the establishment and release of an end-to-end connection, reliable delivery of data packets, flow and congestion control. The reliability mechanisms of VTP must cope with frequent packet losses, reordering, high RTT and high RTT jitter. The performance of VTP mainly depends on its ability to adapt quickly to varying path characteristics.
The key features of VTP are:
• The VTP sender uses a rate-based transmission scheme. The transmission rate is determined by a rate-timer that steadily schedules the transmission of data packets when multi-hop connectivity between source and destination is assumed.
• VTP decouples congestion control from error and flow control, mainly to avoid throughput reduction for packet loss not related to congestion. In VANETs, packet losses are frequent because of high mobility and the result- ing topological changes. These losses must not invoke congestion control. • VTP uses explicit signaling of available bandwidth from intermediate nodes
for congestion control. The estimation of available bandwidth by intermedi- ate nodes uses information from the MAC layer protocol.
• VTP provides reliability via retransmissions of lost packets. Selective ac- knowledgments (SACKs) report lost packets to the VTP sender. The receiver transmits SACKs in dynamic intervals. It adjusts the interval according to the current transmission rate and the source-destination distance.
• The VTP sender uses statistical knowledge to predict the expected com- munication behavior of a connection. In absence of acknowledgments, the expected communication duration for the respective source-destination dis- tance assists the rate timer calculation.
A simulative study evaluates the throughput and fairness of VTP in static and mobile wireless environments and compares these metrics to the performance of TCP.
VTP maintains a constant transmission rate and reacts quickly to disruption or congestion, based on feedback (or absence of feedback) from intermediate nodes. Selective acknowledgments inform the sender about received and missing packets in order to provide reliability by retransmissions. VTP uses statistical knowledge to predict connection behavior, such as expected communication duration, and adapts its transmission rate accordingly.
As a main result, VTP provides reliable end-to-end connections and outper- forms the varying throughput and unfairness of TCP by maintaining a steady through- put above the average throughput of TCP.
The future work will adapt and evaluate VTP in city scenarios, assuming that VTP can achieve similar performance like in highway scenarios when the under- lying routing protocol maintains similar packet delivery ratios. VTP will be im- plemented in the framework of the NoW project and evaluation via measurements will be performed. Finally, interoperability to TCP is required in order to allow access to the Internet or fixed networks at the roadside. This can for example be achieved by installing translation proxies at road side access points or tunnel VTP in TCP over the fixed network.
3.7. SUMMARY AND CONCLUSION 97 The previous chapter designed a vehicular transport protocol for unicast com- munication. Beyond these point-to-point applications, one of the main goals of VANETs is the increase of road safety by reliable point-to-multipoint distribution of safety information to endangered vehicles.
Typically, safety applications require the efficient and reliable distribution of information to vehicles inside a geographically restricted target area over time. The information should be kept alive in the target area for the lifetime of the safety event, i.e., particularly vehicles that enter the target area after the initial message is distributed must be informed.
The following Chapter 4 designs and evaluates a time-extended reliable geo- graphical flooding algorithm that provides reliable and efficient distribution of in- formation in a target area over time.