REDUCTION D1.1 2014 08 31 FINAL - Europa · 2017. 4. 20. · Communication Consortium (C2CCC) which...

REDUCTION 2011‐2014

Deliverable 1.1

Report on the design and architecture of onboard technology and wireless

communication technology 31‐08‐2014

D1.1 [Report on the design and architecture of onboard technology and wireless communication technology]

II

Public Document


III

Project acronym: REDUCTION

Project full title: Reducing Environmental Footprint based on Multi‐Modal Fleet management Systems for Eco‐Routing and Driver Behaviour Adaptation

Work Package: WP1

Document title: Report on the design and architecture of onboard technology and wireless communication technology

Version: 3.0

Official delivery date: 31/08/2014

Actual publication date: 31/08/2014

Type of document: Report

Nature: Public

Authors: Dimitrios Katsaros (UTH), Leandros Tassiulas (UTH), Iordanis Koutsopoulos (UTH), Leandros Maglaras (UTH), Donatos Stavropoulos (UTH)


IV

Approved by: REDUCTION consortium partners

Version Date Sections Affected

1.0 February 26, 2012 Initial version

1.5 February 28, 2012 Review comments processed

2.0 February 12, 2014 Updated to reflect 2nd annual review comments

3.0 August 21, 2014 Minor corrections

Executive Summary Seamless and ubiquitous connectivity requires hardware solutions and communication and networking “algorithmic technology” that will form the “middleware” between the existing infrastructure‐based network technology and the latest advances in wireless technologies and electronics. In this deliverable, we deal with the highly dynamic and volatile environment of vehicular ad hoc networks, and present the appropriate hardware in the form of a DSRC communications box, and a suite of algorithms that will address the scalability issues of the ad hoc network and will also support the communication between vehicles and also the communication between fixed infrastructure elements and vehicles.


5

Table of Contents Executive Summary.................................................................................................................................IV

Table of Contents ....................................................................................................................................... 5

List of Tables ............................................................................................................................................... 6

List of Figures ............................................................................................................................................. 6

Glossary....................................................................................................................................................... 7

1. Introduction ............................................................................................................................................ 8

1.1 Project Overview.............................................................................................................................. 8

1.2 Work Package Objectives and Tasks ............................................................................................. 8

1.3 Objective of this Deliverable........................................................................................................... 9

2. Related Work and REDUCTION....................................................................................................... 10

2.1 A brief history of VANETs ........................................................................................................... 10

2.2 Wireless communications technologies for VANETs ............................................................... 12

3. Framework and Methodology ........................................................................................................... 14

3.1 Wireless communication technology for REDUCTION........................................................... 14

3.1.1 Why DSRC?.............................................................................................................................. 14

3.1.2 What is DSRC?......................................................................................................................... 15

3.2 Communication hardware for REDUCTION ............................................................................ 16

3.3 Ad hoc network’s architecture for REDUCTION...................................................................... 18

3.3.1 Alternative network architectures ........................................................................................ 18


6

3.3.2 A two tier architecture for REDUCTION ............................................................................ 19

3.3.3 Clustering proposal ................................................................................................................ 21

3.3.4 Cross‐layer packet scheduling/routing for the REDUCTION’s architecture ................. 23

3.3.5 Cross‐layer packet scheduling‐routing proposal ............................................................... 23

3.4 Infrastructure‐based communication: The case of I2V information dissemination ............. 24

3.4.1 Background work on broadcast indexing ........................................................................... 25

3.4.2 Broadcast indexing proposal ................................................................................................. 26

3.5 Algorithms for ad hoc routing and broadcast indexing........................................................... 26

3.5.1 The Enhanced Layered Backpressure algorithm................................................................ 26

3.5.2 The DiSAIn indexing algorithm............................................................................................ 28

4. Risk Analysis and lessons learned..................................................................................................... 31

5. Conclusion ............................................................................................................................................ 32

References ................................................................................................................................................. 33

List of Tables Table 1. Summary table for the objectives of the deliverable. ............................................................. 9

Table 2. Wireless technologies versus vehicular applications. .......................................................... 15

List of Figures Figure 1. Generic architecture (to be considered in REDUCTION).................................................. 12

Figure 2. DSRC versus other wireless technologies. ........................................................................... 14


7

Figure 3. Layered architecture for DSRC communication. ............................................................... 15

Figure 4. DSRC band. .............................................................................................................................. 16

Figure 5. A typical DSRC box................................................................................................................. 17

Figure 6. WBSS distribution (from [19])................................................................................................ 17

Figure 7. Architectural categories of decentralized VANETs (from [33]). ....................................... 18

Figure 8. Network with 3 clusters.......................................................................................................... 20

Figure 9. Node 3 has the lowest ΣF and is the clusterhead................................................................ 22

Figure 10. I2V type of communication.................................................................................................. 25

Figure 11. Example network with a moving node. ............................................................................. 27

Figure 12. Impact of moving nodes on the delay performance of BP, LayBP and Enhanced LayBP for the dynamic network with the 3 layers. ............................................................................. 28

Figure 13. A small example of the DiSAIn indexing strategy............................................................ 29

Figure 14. Impact of number of items on mean tuning time. ............................................................ 30

Figure 15. The NITlab wireless testbed and the ICARUS wireless communications box. ............ 31

Glossary I2V Infrastructure‐to‐Vehicle IVC Inter‐vehicular communications OBU On‐Board Unit RSU Road‐Side Unit V2I Vehicle‐to‐Infrastructure V2V Vehicle‐to‐Vehicle VANET Vehicular Ad‐hoc Network


8

1. Introduction

1.1 Project Overview

Reduction of CO2 emissions is a great challenge for the transport sector nowadays. Despite recent progress in vehicle manufacturing and fuel technology, still a significant fraction of CO2 emissions in EU cities is resulting from vehicular transportation. Therefore, additional innovative technologies are needed to address the challenge of reducing emissions. The REDUCTION project focuses on advanced ICT solutions for managing multi‐modal fleets and reducing their environmental footprint. REDUCTION collects historic and real‐time data about driving behavior, routing information, and emissions measurements that are processed by advanced predictive analytics to enable fleets enhancing their current services as follows:

1) Optimizing driving behavior: supporting effective decision making for the enhancement of drivers’ education and the formation of effective policies about optimal traffic operations (speeding, braking, etc.), based on the analytical results over the data that associate driving‐behavior patterns with CO2 emissions;

2) Eco‐routing: suggesting environmental‐friendly routes and allowing multi‐modal lets to reduce their overall mileage automatically; and

3) Support for multi‐modality: offering a transparent way to support multiple transportation modes and enabling co‐modality.

REDUCTION follows an interdisciplinary approach and brings together expertise from several communities. Its innovative, decentralized architecture allows scalability to large fleets by combining both V2V and V2I approaches. Its planned commercial exploitation, based on its proposed cutting edge technology, aims at providing a major breakthrough in the fast growing market of services for ʺgreenʺ fleets in EU and worldwide, and present substantial impact to the challenging environmental goals of EU.

1.2 Work Package Objectives and Tasks

WP1 deals with hardware design and development issues and also with basic wireless communication infrastructure and networking aspects. Its objective is to develop the on‐board technology taking also into account the requirement for supporting multi‐modal fleets for passenger transport.


9

WP1 consists of three tasks:

[T1.1] Design and architecture.

[T1.2] Onboard technology.

[T1.3] VANET packet scheduling/routing and information dissemination.

The present deliverable is the report on T1.1. This task will first collect requirements on the overall architecture of the onboard technology. This is done in close collaboration with tasks T2.1, T3.1, and T4.1. Then, an initial architecture for technical component communication, set‐up and installation routines of devices is implemented. Existing and/or novel communication protocols will be used by the other components to interact with the distributed infrastructure. T1.2 is about onboard technology, i.e., the hardware involved, T1.3 is about VANET packet scheduling‐routing and information dissemination, and finally T1.4 focuses on intelligent V2V and V2I communication.

1.3 Objective of this Deliverable

The goal of this deliverable is to select the appropriate communication technologies for REDUCTION, to decide the architecture at the network level that REDUCTION will be based upon, to recognize which are the topics areas that existing work is not adequate for REDUCTION and set forward efforts for the development of novel algorithms and techniques to address the specific goals that REDUCTION has set. Although the techniques that will be developed here are appropriate for any ad hoc network, we emphasize that there are to be used for vehicular ad hoc networks to serve passenger fleets. Table 1 summarizes the task/report objectives.

Objective to achieve Task Methodology

Wireless communication DSRC‐based suite of standards

DSRC box, UTH’s ICARUS nodes

Ad hoc network scalability Clustering Spring, social‐based clustering

Routing Backpressure LayBP Selective channel tuning for I2V Broadcast indexes DiSAIn

Table 1. Summary table for the objectives of the deliverable.


10

2. Related Work and REDUCTION

2.1 A brief history of VANETs Vehicular Ad Hoc Networks (VANETs) [5],[32] constitute an emerging new technology that integrates the capabilities of new generation wireless networks to vehicles. The idea is to provide: (1) ubiquitous connectivity ‐ while on the road‐ to mobile users, who are otherwise connected to the outside world through other networks at home or at the workplace, and (2) efficient vehicle‐to‐vehicle communications that enable the Intelligent Transportation Systems (ITS). Therefore, vehicular ad hoc networks are also called Inter‐Vehicle Communications (IVC) [28] or Vehicle‐to‐Vehicle (V2V) communications.

The progresses of Intelligent Transport Systems (ITS) have accelerated and utilised recent advances in electronics and computer technologies. ITS are essentially the merger of developments in computing, information technology and telecommunications, coupled to automotive and transportation sector expertise. Vehicles communicate to each other to share some important and emergent information in typical application scenarios. ITS present cost‐effective solutions to a wide range of applications, such as electronic toll collection, emergency vehicle notification systems, automatic road enforcement, or fleet management systems.

VANET and IVC have drawn a significant research interests from both academia and industry. One of the earliest studies on IVC was started by JSK (Association of Electronic Technology for Automobile Traffic and Driving) of Japan in the early 1980s. Later, projects such as the California PATH and Chauffeur of EU have also demonstrated the technique of coupling two or more vehicles together electronically to form a train. Recently, the European project CarTALK 2000 tries to investigate problems related to the safe and comfortable driving based on inter‐vehicle communications. On the other hand, several major automobile manufacturers have already begun to invest in inter‐vehicle networks. Audi, BMW, DaimlerChrysler, Fiat, Renault and Volkswagen have united to create a non‐profit organization called Car2Car Communication Consortium (C2CCC) which is dedicated to the objective of further increasing road traffic safety and efficiency by means of inter‐vehicle communications. IEEE also formed the new IEEE 802.11p task group, which focuses on providing wireless access for the vehicular environment.

VANETs are distinguished from other kinds of ad hoc networks in the following aspects [9],[20]:


11

Highly dynamic topology. Due to high speed of movement between vehicles, the topology of VANETs is always changing. Assuming, e.g., that the wireless transmission range of each vehicle is 250 m, so that there is a link between two cars whose in between distance is less than 250 m. In the worst case, if two cars with the speed of 60 mph (25 m/sec) are driving in opposite directions, the link will last only for at most 10 sec.

Frequently disconnected network. Due to the same reason, the connectivity of the VANETs could also be changed frequently. Especially when the vehicle density is low, it has higher probability that the network is disconnected. One possible solution is to pre‐deploy several relay nodes or access points along the road to keep the connectivity.

Sufficient energy and storage. VANET nodes have ample energy and computing power (including both storage and processing).

Geographical type of communication. Compared to other networks that use unicast or multicast, where the communication end points are defined by ID or group ID, the VANETs often have a new type of communication which addresses geographical areas where packets need to be forwarded.

Mobility modeling and predication. Due to highly mobile node movement and dynamic topology, the factors of mobility model and prediction play an important role in network protocol design for VANETs. Moreover, vehicular nodes are usually constrained by highways, roads and streets, so given the speed and the street map, the future position of the vehicle can be predicted.

Communication environments. VANETs are usually operated in two typical communications environments. In highway traffic scenarios, the environment is relatively simple and straightforward (e.g., one‐dimensional movement); while in city conditions it becomes much more complex. The streets in a city are often separated by buildings, trees and other obstacles. Therefore, there does not always exist a direct line of communications in the direction of intended data communication.

Hard delay constraints. In some VANETs applications, the network does not require high data rates, but it has hard delay constraints. For example, in an automatic highway system, when brake event happens, the message should be transferred and arrived in a certain time to avoid car crash. In this kind of applications, instead of average delay, the maximum delay is crucial.


12

Interaction with on‐board sensors. For example, GPS receivers are increasingly becoming common in cars that help to provide location information for routing purposes. It is assumed that the nodes are equipped with on‐board sensors to provide information that can be used to form communication links and for routing purposes.

2.2 Wireless communications technologies for VANETs Vehicles may utilise a variety of wireless technologies to communicate with other devices, but the dominant is Dedicated Short‐Range Communication (DSRC) [11], which is designed to support a variety of applications based on vehicular communication. Wireless Access in Vehicular Environment (WAVE) is a term used to describe the suite of IEEE P1609.x standards that are focused on MAC and network layers. WAVE is fairly complex and is built over the IEEE 802.11 standards by amending many tweaks to guarantee fast reliable exchange of safety messages. WAVE is the core part of DSRC; however, either of the two terms is commonly used arbitrarily. In some cases, the term DSRC is used as a more general term compared to WAVE.

Figure 1. Generic architecture (to be considered in REDUCTION).

The DSRC network is built over two basic units: Road‐Side Unit (RSU) and On‐Board Unit (OBU). The RSU is, typically, a stationary unit that connects roaming vehicles to the access network,

Cluster leaders

RSU RSU

V2V comm V2I comm


13

which, internally, is connected to a larger infrastructure or to a core network. The OBU is, typically, a network device fixed in a roaming vehicle and is connected to both the DSRC wireless network and to an in‐vehicle network (e.g., the Car‐Area Network ‐ CAN). This generic architecture, which will be considered in REDUCTION, is depicted in Figure 1.

Having presented the basic terminology about the implementation of a VANET, in the rest of this report, we describe the hardware, communication technology and network architecture that comprises the heart of the REDUCTION networking architecture.


14

3. Framework and Methodology

3.1 Wireless communication technology for REDUCTION

3.1.1 Why DSRC? Various communication technologies could be considered for REDUCTION. In Figure 2 (adopted from [19]), we can see the characteristics of DSRC relatively to other wireless communications technologies. It follows that its data rate along with its monetary cost are two appealing factors for its adoption.

Figure 2. DSRC versus other wireless technologies.

Besides of that, DSRC is currently considered the most promising wireless standard that can be used to connect infrastructure (like roadside) to vehicle (I2V) and vehicle‐to‐vehicle (V2V). DSRC standard is based on the WiFi architecture. Relevant application layer consortiums, such as the Vehicle‐Infrastructure Integration (VII now called IntelliDriveSM), Cooperative Intersection Collision Avoidance Systems (CICAS) and others have developed their architect with DSRC services in mind [35]. Table 2 summarizes the reasons for which DSRC is considered as appropriate for most of the vehicle‐based applications.

Collision avoidance DSRC

Road sign notifications DSRC, CALM Safety

Incident management WiFi, DSRC, cellular network

Traffic management DSRC, cellular network, DAB Efficiency

Road monitoring IR, ZigBee, DSRC

Entertainment MMWAVE, WLAN, WiMAX, DVB, DVB‐H Comfort

Contextual information DSRC, cellular network, DAB


15

Table 2. Wireless technologies versus vehicular applications.

3.1.2 What is DSRC? Figure 3 illustrates the protocol stack for DSRC communication, including shorthand names of protocols and standards intended for use at the various layers. At the PHY and MAC layers DSRC utilises IEEE 802.11p Wireless Access for Vehicular Environments (WAVE), a modified version of the familiar IEEE 802.11 (WiFi) standard. In the middle of the stack DSRC employs a suite of standards defined by the IEEE 1609 Working Group: 1609.4 for Channel Switching, 1609.3 for Network Services (including the WAVE Short Message Protocol (WSMP), and 1609.2 for Security Services. DSRC also supports use of well‐known Internet protocols for the Network and Transport layers, i.e., Internet Protocol version 6 (IPv6), User Datagram Protocol (UDP) and Transmission Control Protocol (TCP). These protocols, defined by the Internet Engineering Task Force (IETF), are stable and well documented. The choice between using WSMP or IPv6+UDP/TCP depends on the requirements of a given application. Single‐hop messages, such as those upon which collision prevention applications are based, typically use the bandwidth‐efficient WSMP, while multi‐hop packets use IPv6 for its routing capability.

Figure 3. Layered architecture for DSRC communication.

Detailed information of these layers can be found at [11]. DSRC has been allocated the spectrum from 850 GHz to 5.925 GHz, i.e., the “5.9 GHz band”. This spectrum is divided into seven, 10


16

MHz channels with a 5‐MHz guard band at the low end, as illustrated in Figure 4. Pairs of 10 MHz channels can also be combined into a 20 MHz channel.

Figure 4. DSRC band.

3.2 Communication hardware for REDUCTION A typical DSRC box – see Figure 5 for an example – hosts a powerful INTEL multi‐core processor ready to take on any task ranging from vision, sensor fusion and of course V2X. Performance to bear whilst keeping a small form‐factor applicable in an automotive environment was an important design factor as well. The device is complemented with a robust Linux OS and ready to be deployed in current V2X projects. Typical features of it include:

• standard x86 architecture INTEL® CORE 2 DUO 2 GHz with temp. range 0°C to ~60°C, • onboard 2GB DDR2 RAM, • onboard 2/4/8 GB Solid State Disk, • ETH 100 MBit • USB2.0 • CAN • IEEE 802.11p radio • wide‐range power supply (8 ~ 32V, 20W) • customized enclosure for enhanced heat dissipation • automotive grade connectors • customized & robust Linux OS


17

Figure 5. A typical DSRC box.

The wireless connection between RSU and OBU is based on WAVE standards suite. As OBUs move between communication zones, vehicles exchange information with the roadside; in addition, vehicles use the same WAVE media to communicate with each other.

Figure 6 presents a plan view where the light posts shown off the green are equipped with RSUs. Each RSU has a communication zone indicated by the triangular conem, and vehicles go through different communication zones as they drive, e.g., on a highway. To define different WAVE communication zones, we can consider the term WAVE Basic Service Set (WBSS) as a unique identifier for each communication zone. Vehicles must associate with only one WBSS at a time. Hence, each communication zone has its own WBSS. Figure 6 exemplifies the first communication zone from the left is WBSS‐1 and the last is WBSS‐4. Vehicles close to each other, such Car‐B and Car‐C, may have a V2V communications, such as WBSS‐5. WBSS‐1 is an I2V, therefore vehicles may be part of either I2V V2V session at the same time. The communication zone covered by each IEEE 802.11p RSU is limited to a maximum of 1 Km diameter and uses 5.9 GHz radio transmission. OBUs are expected to join the WBSS of the closest RSU, exchange information, and typically leave within limited time (mean estimate 3.6 sec).

Figure 6. WBSS distribution (from [19]).


18

3.3 Ad hoc network’s architecture for REDUCTION Several decentralized traffic information systems have been proposed in the past to reduce the maintenance costs of the centralized server approach. These can be categorized into four families that are detailed in the following [33].

Figure 7. Architectural categories of decentralized VANETs (from [33]).

3.3.1 Alternative network architectures

Single‐tier VANETs

In this type, depicted in Figure 7a, we have a flat vehicular ad hoc network, where nodes (i.e., vehicles) communicate with each other via their OBUs. Vehicles communicate with each other


19

through inter‐vehicle communication, and periodically broadcast their current speeds and positions to neighboring vehicles. A part of the traffic information, a vehicle receives may also be propagated to its neighbors through broadcast messages. Based on the received messages, a vehicle can generate traffic reports.

Single‐tier p2p over VANETs

The above architecture can be further extended to a Peer‐to‐Peer (P2P) over VANET architecture, as shown in Figure 7b. Vehicles form an application‐layer P2P overlay network on top of the VANET. The P2P overlay can be unstructured, such as Gnutella, or structured, such as Chord. The vehicles share their resources (i.e., traffic information) and retrieve resources from others through the P2P overlay. The application‐layer P2P overlay communication relies on the routing protocol of the underlying VANET. A vehicle should establish a routing path in the VANET first; then an application‐layer message can be transmitted along the route to another vehicle.

Single‐tier infrastructure‐based p2p

Another single‐tier architecture involves forming a P2P overlay through an infrastructure network, as shown in Figure 7c. Vehicles are required to have a broadband wireless interface to access the infrastructure network. Vehicles communicate with each other through infrastructure communication instead of ad‐hoc communication.

Two‐tier VANETs

Vehicles are first organized into groups in VANETs. Similar to the single‐tier VANET architecture, traffic information is broadcasted and exchanged among vehicles through IVC. Some vehicles in the groups are selected to form a high‐tier P2P overlay through infrastructure wireless communication. These vehicles are called superpeers (or cluster heads ‐ CHs) and serve as a bridge between the high‐tier and low‐tier networks to handle message exchanges and lookups. This architecture (Figure 7d) is the most promising one, since it is scalable and, thus, the most appropriate for the high velocity of the moving vehicles.

3.3.2 A two tier architecture for REDUCTION REDUCTION will be based on vehicle clustering. Clustering is the process of separating the nodes of a network into organized partitions called clusters. The clusters form sub‐networks in the overall network, thus forming the hierarchical topology. Figure 8 illustrates a clustered ad hoc network. Nodes in a cluster must be one of the following types:

• Clusterhead (CH): An elected node that acts as the local controller for the cluster. The


20

clusterheadʹs responsibilities may include: routing, relaying of inter‐cluster traffic from cluster members, scheduling of intra‐cluster traffic, and channel assignment for cluster members.

• Cluster Member: A normal node belonging to a cluster. Cluster members usually do not participate in routing, and they are not involved in inter‐cluster communication.

• Gateway Node: This is an optional node, which is used in some clustering schemes. The gateway node belongs to more than one cluster, acting as the bridge between clusterheads. When present, the gateway nodes participate in both forwarding of inter‐cluster traffic and the routing process. The clusterheads and gateway nodes form the backbone network.

Figure 8. Network with 3 clusters.

The recent research in cluster‐based MAC and routing schemes for Vehicle Ad Hoc Networks (VANETs) motivates the necessity for a stable VANET clustering algorithm. Due to the highly mobile nature of VANETs, mobility must play an integral role in cluster formation. By clustering the vehicles into groups of similar mobility, the relative mobility between communicating neighbor nodes can be reduced. Both delay‐intolerant (e.g., safety messages) and delay‐tolerant (e.g. road/weather information) data will need to be transmitted, necessitating Quality‐of‐Service (QoS) requirements. Although there is not a VANET clustering scheme focused on cluster stability, there are many mobility‐based clustering techniques for ad hoc networks. A well‐known mobility‐based clustering technique is MOBIC [1], which is an extension of the Lowest‐ID algorithm [13]. In Lowest‐ID, each node is assigned a unique ID, and the node with the lowest ID in its two‐hop neighborhood is elected to be the cluster head. In MOBIC, an aggregate local mobility metric is the basis for cluster formation instead of node ID. Other schemes are based on the idea of exploiting the RSUs [2], the mobility of vehicles [25],[29], their position [34], the multi‐channel feature of DSRC [40], or the specific road


21

topology where the vehicles are roaming [23].

In [17] a random‐walk‐based layering‐clustering is developed that is appropriate for static networks, but it is centralized. The algorithm runs in blocks where in each block a node i is selected as a source. The algorithm takes successive random steps from node i for a predefined number of steps h for K iterations. Every node holds a counter indicating how many times it is visited from the random walk algorithm in the current block. The procedure moves in steps, inserting at every iteration at most two nodes in the possible layer, leading to a relatively fair clustering of the nodes after its termination. This procedure is repeated for each node belonging to a temporary layer until all the layers are fixed.

In VANETs, dynamic topology causes routing difficulties as well as congestion from flooding, and the dense network leads to the hidden terminal problem. A clustered structure can make the network appear smaller and more stable in the view of each node, and moreover can address is the “broadcast storm problem” [36]. The broadcast storm problem describes the congestion resulting from rebroadcasts and flooding in an ad hoc network. The dynamic topology of VANETs demands a high frequency of broadcast messages to keep the surrounding vehicles updated on position and safety information. In addition, many routing algorithms necessitate flooding the network to find routes, which in a dynamic network needs to be done frequently to keep routes updated. All of this flooding leads to severe congestion, which can be alleviated by a clustered topology. When the network is clustered, only the clusterhead participates in finding routes, which greatly reduces the number of necessary broadcasts. In addition, MAC schemes can greatly reduce interference.

3.3.3 Clustering proposal Even though the two‐tier clustering schemes have established themselves as the preferred choice, nevertheless none of those proposed so far is the clear winner. Therefore, we are investigating the idea of “spring‐clustering”. The idea is based on force‐directed algorithms. The force‐directed assign forces among the set of edges and the set of nodes; the most straightforward method is to assign forces as if the edges were springs and the nodes were electrically charged particles. The entire graph is then simulated as if it were a physical system. The forces are applied to the nodes, pulling them closer together or pushing them further apart.

The spring‐clustering method considers a spring‐like force for every pair of nodes (i,j) where the ideal length δij of each spring is proportional to the graph‐theoretic distance between nodes i and j. All the links among nodes are treated as springs with force Fij.


22

ijijij xkF = (1),

where xij is the Euclidean distance between the nodes and kij is computed using the following equation:

same thestays distance the,1closer moving are nodes the,2

away moving are nodes twothe,1−=ijK

Each node has knowledge about its current position, the position of each other connected to it node, and the velocity vectors of itself and the other nodes. We describe our algorithm in the following steps:

1 All nodes send and receive “Hello” messages to/from their neighbors. Each node calculates the pairwise relative force F using equation (1).

2 Each node computes the average force Σf from its 1‐hop neighbors.

3 If a node has the lowest value of Σf (aggregate forces) amongst all its neighbors, it assumes the status of a Cluster Head; for an example, see Figure 9. Otherwise, it declares itself to be a Cluster Member. If a node is a neighbor of two clusterheads, then it becomes a “gateway” node.

Figure 9. Node 3 has the lowest ΣF and is the clusterhead.


23

3.3.4 Cross‐layer packet scheduling/routing for the REDUCTION’s architecture Routing in VANETs [30] can be classified into five categories as follows:

• Ad hoc. VANETs and MANETs share the same principle: not relying on fixed infrastructure for communication, and have many similarities, e.g., self‐organization, self‐management, low bandwidth and short radio transmission range. Thus, most ad hoc routing protocols are still applicable, such as AODV (Ad‐hoc On‐demand Distance Vector) and DSR (Dynamic Source Routing). AODV and DSR are designed for general purpose mobile ad hoc networks and do not maintain routes unless they are needed.

• Position‐based. Node movement in VANETs is usually restricted only bidirectional movements constrained along roads and streets. So, routing strategies that use geographical location information obtained from street maps, traffic models, or even more prevalent navigational systems on‐board the vehicles make sense [18]. Therefore, geographic routing (position‐based routing) has been identified as a more promising routing paradigm for VANETs [4].

• Cluster‐based. In cluster‐based routing, a virtual network infrastructure must be created through the clustering of nodes in order to provide scalability.

• Broadcast. Broadcast is a frequently used routing method in VANETs, such as sharing traffic, weather, emergency, road condition among vehicles, and delivering advertisements and announcements. Broadcast is also used in unicast routing protocols (routing discovery phase) to find an efficient route to the destination. When the message needs to be disseminated to the vehicles beyond the transmission range, multi‐hop is used. The simplest way to implement a broadcast service is flooding, in which each node re‐broadcasts messages to all of its neighbors except the one it got this message from.

• Geocast routing. Geocast routing [14] is basically a location‐based multicast routing. The objective of a geocast routing is to deliver the packet from a source node to all other nodes with a specified geographical region (Zone of Relevance, ZOR).

A qualitative comparison of these protocols can be found at [12]. Though, none of these protocols offers QoS (e.g., throughput, delay) guarantees.

3.3.5 Cross‐layer packet scheduling‐routing proposal On the other hand, in the world of ad hoc networking the issue of cross‐layer design of communication protocols with throughput optimality has received a lot of attention. The


24

development of a throughput‐optimal routing algorithm for packet radio networks, which is also robust to topology changes, was first presented in [31]. It is based on the Lyapunov drift theory, and it is known as the backpressure (BP) packet scheduling algorithm. Subsequently, the original concept spawns several lines of research in the topic.

The performance of backpressure deteriorates in conditions of low, and even of moderate, load in the network, since the packets “circulate” in the network, i.e., the backpressure algorithm stabilizes the system using all possible paths throughout the network. The net effect of this mechanism is to increase delay. Routing‐loop formation is another drawback of backpressure routing. In many real time applications, such as in cases of voice and video, high end‐to‐end packet delay is unacceptable. Often in such applications, a packet received with high delay is no better than packet loss. We could prevent high end‐to‐end delay in backpressure routing by not forwarding the packets on longer paths. But we still want to maintain the sufficient routes for any source‐destination pair to provide adequate load balancing in case of high traffic load. Generally, these two objectives conflict with each other because few short routes exist in the network.

This cross‐layering approach is very significant for VANETs since it can guarantee throughput optimality, provided that it has acceptable delay performance. The Enhanced Layered Backpressure algorithm (see later section) is such an algorithm with the only drawback that currently runs in a centralized fashion; it requires a network controller that has full knowledge of the network status, i.e., topology and queuing information.

3.4 Infrastructure‐based communication: The case of I2V information dissemination

Apart from PHY and MAC layer issues in the I2V communication between RSUs and vehicles, the task of broadcasting is a fundamental operation that exploits the shared medium, i.e., the wireless channel, to transmit information to clients (vehicles). The communication protocols in such mode can be implemented either as pure pull‐based, pure push‐based or on‐demand broadcasting. In pure pull‐based broadcast, the clients request information via an uplink channel, and subsequently the server allocates a channel for the requesting client, and transmits the information. In pure push‐based broadcast, the server sends information over common broadcast channel to all listening ‘consumers’ (clients). In on‐demand broadcast, the clients pose requests to the server, and the server broadcasts the responses via a shared broadcast channel – thus a single response can satisfy multiple requests. Apparently, (pure‐push and on‐demand) broadcasting is a preferred choice for modern wireless networks, since it overcomes the


25

scalability issues associated with the large number of consumers and the large volume of transmitted data.

Consider, for instance, a data dissemination application in an Infrastructure‐to‐Vehicle (I2V) case. In this push‐based broadcast system, the RSUs broadcast information concerning issues relevant to the moving vehicles, e.g., traffic congestion reports, updated routing instructions for the vehicles of a fleet, and so on. Each RSU constructs a broadcast program with the needed info and broadcasts it periodically. All vehicles can tune in to the broadcast channel in order to retrieve the information (data items) without sending explicit requests for them, thus avoid ’choking’ the uplink channel. Figure 10 illustrates such a scenario. A vehicle wants to retrieve data item B from a broadcast channel. The importance of knowing when item B will be broadcasted is crucial, because the vehicle can decide to accelerate to get it from the next RSU, or slow down to get it from the current RSU. The presence of air indexes would give an answer about the time of broadcast. Regulation of the vehicle’s velocity and knowledge of the distance between successive RSUs could provide a suggestion to the driver about its driving policy.

Figure 10. I2V type of communication.

3.4.1 Background work on broadcast indexing An adaptation of the idea of B+‐tree indexing in wireless environments was first described in [7], where instead of the disk addresses the leaves of the B+‐tree store the arrival time of each datum in the broadcast channel. Similarly, an adaptation of the traditional hash‐based indexing technique in wireless environments was earlier described in [6], and later was generalized in [37]. Hybrids between the two approaches are described in [24], [38]. Adaptation of such indexing schemes (e.g., B+‐tree) to work in multiple broadcast channels is described in [39].


26

They do not propose new schemes but simply different allocation methods for the nodes of the indexing tree.

In all these works it is assumed that: a) there is a global ordering among the transmitted data, and b) the access pattern is uniform. Deviating from the uniform access probability assumption, several works considered the effect of access skew on the design of indexing schemes. A new scheme is proposed in [26], which is a k‐ary version of the basic binary Alphabetic Tree over the data, whereas [39] adapted the indexing method of [7] to deal with non‐uniformity in access. Various methods were based on the construction of a binary or k‐ary Alphabetic Tree to develop indexing schemes for multiple broadcast channels [8], [41]. These methods do not provide new types of tree‐structured indices, but rather a new allocation method for the tree‐structured method of Alphabetic trees to the multiple channels. All these works assume that: a) there is a global ordering among the transmitted data. There are only a couple of works [3], [10], which deviated from both the uniformity and global ordering assumptions.

3.4.2 Broadcast indexing proposal In the context of efficient broadcast indexing, the DiSAIn index is proposed to address the problem of skewed access probabilities observed in real life applications and the appropriate data dissemination in VANETs.

3.5 Algorithms for ad hoc routing and broadcast indexing

3.5.1 The Enhanced Layered Backpressure algorithm

Backpressure [31] is a joint scheduling and routing policy which favors traffic with high backlog differentials. The backpressure algorithm performs the following actions for routing and scheduling decisions at every time slot t:

Resource allocation. For each link assign a temporary weight according to the differential backlog of every commodity.

Scheduling. The network controller chooses the control action that maximizes the sum of weights.

Routing. Backpressure algorithm is throughput‐optimal and discourages transmitting to congested nodes, utilising all possible paths between source and destination. This property, leads to unnecessary end‐to‐end delay when the traffic load is light. Moreover, using longer paths in cases of light or moderate traffic wastes network resources.


27

Implementation. In [17], we developed a delay‐efficient backpressure algorithm based on the creation of layers in the network called LayBP. The main idea is to split the network into layers according to the connectivity among them, which also (usually) implies geographic proximity, as well. These layers divide the initial graph to k smaller networks. Then the algorithm forwards packets according to the destination layer ID, thus effectively reducing the long paths. In some sense, these layers play the role of attractors that attract the packets destined for them and then “disallow” the packets to leave the layer.

Routing protocols must be dynamic in order to cope with mobility of nodes in modern wireless networks. Widely varying mobility characteristics are expected to have a significant impact on the performance of routing protocols that are based on node grouping in order to route packets even if links among nodes are updated. In case of grouping‐based routing protocols, high mobility of nodes, which lead them to change groups, degrades the performance of the methods since this ‘wrong’ information is used in the routing procedure. Although LayBP doesn’t use gateways, it still suffers from this behavior if the layer that the moving nodes belong to, are not updated. The differential backlog of each link is computed according to the difference between current and destination’s node layer. It is clear that LayBP behavior can be affected of ‘misplaced’ nodes. In this case packet may be forwarded to layers different than the desired making the method inappropriate.

In order to cope with node mobility, we designed the Enhanced LayBP algorithm [16]; it is the original LayBP algorithm with one more step, in which moving nodes recalculate their cluster according to their neighborhood. In the initiation, phase every node has a counter for every layer ID, indicating how many neighbors belong to it and a variable indicating the layer that node n belongs to. Every t time slots every moving node update the layer it belongs to according to a dedicated algorithm (for an example, see Figure 11).

13 13 13

Layer 2

3

2 9 10

12118

1

4 6

5 7Layer 1

T_0: (time slot 0) T_1: (time slot 1000)

Layer 3

T_2: (time slot 2000)

Figure 11. Example network with a moving node.


28

Despite its desirable properties, including being throughput optimal and operating on instantaneous queue and channel states without requiring knowledge of the underlying statistical distributions, max‐weight scheduling (Backpressure based routing algorithms) does require global knowledge of network state information, an unrealistic assumption for most real‐world implementations. In practical networks, possessing full system state information requires dissemination of such information to all nodes. Exchanging this network state information leads to increases in protocol complexity and overhead, and, therefore, may result in a reduction in efficiency.

In order to fully implement the Enhanced Layered Backpressure in VANETs, the protocol will be implemented in a distributed fashion along the lines of [21], [22] and [27].

Evaluation. A preliminary simulation‐based evaluation of the proposed Enhanced LayBP algorithm, shows that it is a delay‐aware and throughput‐optimal backpressure algorithm; see, for instance, Figure 12.

Figure 12. Impact of moving nodes on the delay performance of BP, LayBP and Enhanced LayBP for the dynamic network with the 3 layers.

3.5.2 The DiSAIn indexing algorithm To overcome that problem of ‘blindly’ searching among all transmitted data from a RSU, the RSU usually interleaves along with the data, and some index packets, to help the consumers ‘move’ into the broadcasted information. The index packets contain the necessary information to guide the user through the transmitted data, until the user reaches the information. These index packets play the role of the indices that we encounter into the traditional disk‐based database systems. The difference between the two is that the broadcast channel is a one‐


29

dimensional medium, and the interleaving of index packets increases the access time, in an attempt to decrease the tuning time. These performance measures are typically used to measure the efficiency of a wireless data broadcast system; access time is the time elapsed from the moment a request is issued by a client to the moment the requested data is returned, and tuning time is the duration of time a client stays tuned in to collect requested data items. All indexing schemes attempt to achieve the best tradeoff between tuning time and access time.

Implementation. The objective of the DiSAIn index [15] is to address the problem of skewed access probabilities observed in real life applications and the appropriate data dissemination in VANETs. To achieve this, it adds more ‘pointers’ to the original ExpIn index [37], but it retains some of the properties of the ExpIn index: each bucket of the broadcast cycle contains a data bucket and an index table. The index table consists of i entries. Each entry indexes a range of buckets that are 2i−1 to 2i − 1 buckets away and holds the maxkey value of these buckets. But for the last bucket, the DiSAIn maintains another index, the skewed index SI, which points to the most popular element MP of the segment. The construction of the index table takes place in an initiation phase where the distance of MP from the maxkey value of the last segment is calculated for each data element. In case where all data elements of the last segment have equal access probabilities, then the DiSAIn indexes one of these elements at random.

Figure 13. A small example of the DiSAIn indexing strategy.

Figure 13 illustrates a sample DiSAIn index; it is supposed that the elements ‘A’ to ‘H’ are to be indexed, and that the element ‘F’ is the most popular one. In general, one fourth of the total number of pointers point to this ‘hot’ element. If the client tunes into the broadcast channel just before item ‘A’, then it retrieves the index table that corresponds to the bucket of ‘A’. In case where the client issues a query for item ‘F’, a pointer points to it directly, minimizing the tuning time while keeping the access latency constant.


30

Evaluation. A preliminary simulation‐based evaluation of the proposed algorithm against the state‐of‐the‐art method [37], shows that DISAIN is a high performance indexing structure; see, for instance, Figure 14.

Figure 14. Impact of number of items on mean tuning time.


31

4. Risk Analysis and lessons learned

Concerning the risks faced by this task, we would consider them along two dimensions. The first dimension is whether our communications and networking technology are the appropriate ones for REDUCTION; the second dimension is whether the communications hardware will be ready in early enough time I order to investigate the V2V, V2I and I2V communication and networking primitives that we will develop. As far as the first dimensional is concerned, we need to emphasize that since we follow international standards, for which there is sufficient documentation and experience not only in the international community but also to the REDUCTION partners (DELPHI and UTH), we feel confident that the risks are practically non‐existing. Concerning the timely development of hardware, we have to admit that this is a tough issue and we may not follow the planned schedule in a strict manner. Nevertheless, we can use established simulation tools such as ns3, OMNET++, to test the developed algorithms. Moreover, the UTH team has developed its one infrastructure (a wireless testbed) and communication hardware that will be enhanced in the context of REDUCTION and will be used for the REDUCTION needs. Our wireless testbed, namely NITlab and communication box, namely ICARUS node are illustrated in Figure 15.

Figure 15. The NITlab wireless testbed and the ICARUS wireless communications box.


32

5. Conclusion

In this section, we summarize the decisions that will mostly affect the design and architecture of REDUCTION’s system, along with some significant open issues relevant to its overall architecture.

Wireless communications technology. DSRC technology in the 5.9 GHz band has the potential to support many different types of applications, including collision avoidance applications, traffic regulation, passenger fleet management that can save tens of thousands of lives and billions of dollars in the EU. This technology depends fundamentally on standards‐based interoperability, and it is the preferred choice for REDUCTION’s wireless communication technology, as well as for many projects completed or currently running [9] in EU, USA, and Japan. There is no doubt that this is the wireless technology that should be used.

Network architecture. A two‐tier (clustered) architecture is scalable and it achieves much higher lookup success rates than flat VANET systems and outperforms single‐tier infrastructure‐based systems in terms of success rate, latency, and maintenance cost. It is the only approach to address the broadcast storm problem. The open issues for the two‐tier architecture are cluster‐head selection and redundancy, stability of the clustering mechanism and fast re‐clustering operations. Therefore, we need to proposed new clustering algorithms that will consider characteristics of the vehicles (e.g., size, velocity) as well as the “social” behavior of their drivers.

Routing and information dissemination. Backpressure‐type algorithms are the most promising ones for supporting throughput‐optimal packet scheduling/routing for ad hoc communications and can also be used for V2I communication. However, they suffer from delay problems and are centralized. Therefore, the open issues are to develop delay‐aware backpressure algorithms along the lines of [17], and also to transform them into distributed protocols.

As long as the I2V type of communication is concerned, even though there is significant experience in the communications/networking community since cellular, WiMAX and WLANs networks are operating for years, there are some issues, such as fast access to the transmitted information, that need further investigation. Towards this direction, we will develop air indexing schemes to support fast, selective tuning to the broadcast channel.

Overall, we are confident that the present report has achieved the goals set by task T1.1.


33

References [1] P. Basu, N. Khan, and T. Little, “A mobility based metric for clustering in mobile ad hoc networks”,

Proceedings of the IEEE International Conference on Distributed Computing Systems Workshop (ICDCSW), pp. 413–418, 2001.

[2] W. Chen and S. Cai, “Ad hoc peer-to-peer network architecture for vehicle safety communications”, IEEE Communications magazine, pp. 100-107, April, 2005.

[3] M.-S. Chen, K.-L. Wu, and P. Yu, “Optimizing index allocation for sequential data broadcasting in wireless mobile computing”, IEEE Transactions on Knowledge and Data Engineering, vol. 15, no. 1, pp. 161–173, 2003.

[4] R.J. Hall, “An improved geocast for mobile ad hoc networks”, IEEE Transactions on Mobile Computing, vol. 10, no. 2, pp. 254-266, 2011.

[5] H. Hartenstein and K.P. Laberteaux, “A tutorial survey on vehicular ad hoc networks”, IEEE Communications magazine, pp. 164-171, June 2008.

[6] T. Imielinski, S. Viswanathan, and B. R. Badrinath, “Power efficient filtering of data on air”, Proceedings of the Conference on Extending Data Base Technology (EDBT), 1994, pp. 245–258.

[7] T. Imielinski, S. Viswanathan, and B. R. Badrinath, “Data on air: Organization and access”, IEEE Transactions on Knowledge and Data Engineering, vol. 9, no. 3, pp. 353–372, 1997.

[8] S. Jung, B. Lee, and S. Pramanik, “A tree-structured index allocation mathod with replication over multiple broadcast channels in wireless environments”, IEEE Transactions on Knowledge and Data Engineering, vol. 17, no. 3, pp. 311–325, 2005.

[9] G. Karagiannis, O. Altintas, E. Ekici, G. Heijenk, B. Jarupan, K. Lin and T. Weil, “Vehicular networking: A survey and tutorial on requirements, architectures, challenges, standards and solutions”, IEEE Communications Surveys & Tutorials, vol. 13, no. 4, pp. 584-616, 2011.

[10] D. Katsaros, N. Dimokas, and Y. Manolopoulos, “Generalized indexing for energy-efficient access to partially ordered broadcast data in wireless networks”, Proceedings of the IEEE International Database Engineering and Applications Symposium (IDEAS), pp. 89–96, 2006.

[11] J.B. Kenney, “Dedicated Short-Range Communications (DSRC) standards in the United States”, Proceedings of the IEEE, vol. 99, no. 7, 2011.

[12] F. Li and Y. Wang, “Routing in vehicular ad hoc networks: A survey”, IEEE Vehicular Technology magazine, pp. 12-22, June, 2007.

[13] C. Lin and M. Gerla, “Adaptive clustering for mobile wireless networks”, IEEE Journal on Selected Areas in Communications, vol. 15, no. 7, pp. 1265–1275, Sep 1997.

[14] C. Maihofer, “A survey of geocast routing protocols”, IEEE Communications Surveys & Tutorials, vol. 6, no. 2, pp. 32-42, 2004.

[15] L. Maglaras and Dimitrios Katsaros, “Distributed skip air index for smart broadcasting in intelligent transportation systems”, Proceedings of the IEEE Intelligent Vehicles Symposium, June, 2012.

[16] L. Maglaras and Dimitrios Katsaros, “Delay-efficient opportunistic routing in wireless ad hoc networks”, submitted for journal publication.

[17] L. Maglaras and Dimitrios Katsaros. "Layered backpressure scheduling for delay reduction in ad hoc networks", Proceedings of the 12th IEEE International Symposium on a World of wireless, Mobile and Multimedia Networks (WoWMoM), 2011.

[18] M. Mauve, J. Widmer and H. Hartenstein, “A survey on position-based routing in mobile ad hoc networks”, IEEE Network magazine, pp. 30-39, 2001.

[19] Y.L. Morgan, “Notes on DSRC & WAVE standards suite: Its architecture, design, and characteristics”, IEEE Communications Surveys & Tutorials, vol. 12, no. 4, pp. 504–518, 2010.

[20] G.Pallis, D. Katsaros, M.D. Dikaiakos, N. Loulloudes and L. Tassiulas, “On the structure and


34

evolution of vehicular networks”, Proceedings of the IEEE/ACM International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS), pp. 502-511, 2009.

[21] S.T. Rager, E.N. Ciftcioglu, A. Yener, T.F. La Porta and Michael J. Neely, “Distributed backpressure protocols with limited state feedback”, Proceedings of the IEEE Conference on Military Communications (MILCOM), 2011.

[22] J. Ryu, V. Bhargava, N. Paine and S. Shakkottai, “Back-pressure routing and rate control of ICNs”, Proceedings of the IEEE/ACM International Conference on Mobile Computing and Networking (MobiCom), pp. 365-376, 2010.

[23] E. Sakhaee and A. Jamalipour, “Stable clustering and communications in pseudolinear highly mobile ad hoc networks”, IEEE Transactions on Vehicular Technology, vol. 57, no. 6, pp. 3769-3777, 2008.

[24] A. Seifert and J.-J. Hung, “FlexInd: A flexible and parameterizable air indexing scheme for data broadcast systems”, Proceedings of the Conference on Extending Data Base Technology (EDBT), ser. Lecture Notes in Computer Science, vol. 3820, pp. 902–920. 2006.

[25] C. Shea, B. Hassanabadi and S. Valaee, “Mobility-based clustering in VANETs using affinity propagation”, Proceedings of the IEEE Global Communications Conference (GLOBECOM), 2009.

[26] N. Shivakumar and S. Venkatasubramanian, “Efficient indexing for broadcast based wireless systems”, ACM/Baltzer Mobile Networks and Applications, vol. 1, no. 4, pp. 433–446, 1996.

[27] Umesh Kumar Shukla, “Backpressure policies for wireless ad hoc networks”. Master Thesis. [28] M. Sichitiu et al., “Inter-vehicle communications systems: A survey”, IEEE Communications

Surveys & Tutorials, vol. 10, no. 2, pp. 88-105, 2008. [29] E. de Souza, O. Nikolaidis and P. Gburzynski, “A new aggragete local mobility (ALM) clustering

algorithms for VANETs”, Proceedings of the IEEE Conference on Communications (ICC), 2010. [30] C. Suthaputchakun and Z. Sun, “Routing protocol in intervehicle communication systems: A survey”,

IEEE Communications magazine, vol. 49, no. 12, pp. 150-156, 2011. [31] L. Tassiulas and A. Ephremides, “Stability properties of constrained queueing systems and

scheduling policies for maximum throughput in multihop radio networks”, IEEE Transactions on Automatic Control, vol. 37, no. 12, pp. 1936–1948, 1992.

[32] Y. Toor et al., “Vehicle ad hoc networks: Applications and related technical issues”, IEEE Communications Surveys & Tutorials, vol. 10, no. 3, pp. 74-88, 2008.

[33] S.L. Tsao and C.-M. Cheng, “Design and evaluation of a two-tier peer-to-peer traffic information system”, IEEE Communications magazine, pp. 165-712, May, 2011.

[34] Z. Wang, L. Liu, M.C. Zhu and N. Ansari, “A position-based clustering technique for ad hoc intervehicle communication”, IEEE Transactions on Systems, Man, and Cybernetics – Part C : Applications and Reviews, vol. 38, no. 2, pp. 201-208, 2008.

[35] T. Willke, P. Tientrakool, and N. Maxemchuk, “A survey of inter-vehicle communication protocols and their applications”, IEEE Communications Surveys & Tutorials, vol. 11, no. 2, 2009.

[36] N. Wisitpongphan, O.K. Tonguz, J.S. Parikh, P. Mudalige, F. Bai, and V. Sadekar, “Broadcast storm mitigation techniques in vehicular ad hoc networks”, IEEE Wireless Communications magazine, pp. 84-94, December, 2007.

[37] J. Xu, W.-C. Lee, X. Tang, Q. Gao, and S. Li, “An error-resilient and tunable distributed indexing scheme for wireless data broadcast”, IEEE Transactions on Knowledge and Data Engineering, vol. 18, no. 3, pp. 392–404, 2006.

[38] X. Yang and A. Bouguettaya, “Adaptive data access in broadcast-based wireless environments”, IEEE Transactions on Knowledge and Data Engineering, vol. 17, no. 3, pp. 326–338, 2005.


35

[39] W. G. Yee and S. B. Navathe, “Efficient data access to multi-channel broadcast programs”, Proceedings of the ACM Conference on Information and Knowledge Management (CIKM), pp. 153–160, 2003.

[40] X. Zhang, H. Su, and H.-H. Chen, “Cluster-based multi-channel communications protocols in vehicle ad hoc networks”, IEEE Wireless Communications magazine, pp. 44-51, October, 2006.

[41] J. Zhong, W. Wu, Y. Shi, and X. Gao, “Energy-efficient tree-based indexing schemes for information retrieval in wireless data broadcast”, Proceedings of the International Conference Database Systems for Advanced Applications (DASFAA), pp. 335–351, 2011.

REDUCTION D1.1 2014 08 31 FINAL - Europa · 2017. 4. 20. · Communication Consortium (C2CCC) which...

Documents

Transcript of REDUCTION D1.1 2014 08 31 FINAL - Europa · 2017. 4. 20. · Communication Consortium (C2CCC) which...