Anatomy of Connected Carrss

,2,3

(1.UCLA?MPI Joint Research Laboratory in Ubiquitous Computing，University of California，Los Angeles，California，USA;

2.Universite’Pierre and Marie Curie-LIP6，Paris，France;

3.Computing/Computer Studies Program，Macao Polytechnic Institute，Macao，China)

Anatomy of Connected Carrss

Mario Gerla1,Giovanni Pau1,2,and Rita Tse1,3

(1.UCLA?MPI Joint Research Laboratory in Ubiquitous Computing，University of California，Los Angeles，California，USA;

2.Universite’Pierre and Marie Curie-LIP6，Paris，France;

3.Computing/Computer Studies Program，Macao Polytechnic Institute，Macao，China)

The National Highway Traffic Safety Administration in the US and the European Commission are drafting a regulatory framework that will make the goal of connected vehicles possible by 2020.Control，embedded systems，and communication technologies have devel?oped over the past 10 plus years and are approaching maturity.These will spark a revolution in how we approach driving.Cars will no longer need human drivers;they will be connected and exchange information about navigation，road hazards，traffic conditions，and safety.Travelers will be connected more than ever.Today’s car will become tomorrow’s office and the act of driving will be?come a leisure activity rather than a necessity.The emerging Internet of Vehicle enables application scenarios unimaginable just few years ago.The main challenges are Internet access spectrum scarcity，mobility，intermittent connectivity and scalability.In this arti?cle，we discuss the evolution from intelligent vehicle grid to autonomous，internet?connected vehicles and vehicular cloud.

connected vehicles;Internet of Vehicles;name data networking

1 Introduction

Traditionally，a vehicle has been considered an exten?sion of human ambulation and is easily controlled by the driver’s commands.Recent advances in com?munications，controls and embedded systems have changed this model，paving the way for the intelligent vehicle grid.The car is now a formidable sensor platform，absorbing in?formation from the environment(and from other cars)and feed?ing it to drivers and infrastructure in order to assist with safe navigation，pollution control，and traffic management.The next step in this evolution is just around the corner:the Internet of Connected Autonomous Vehicles.Pioneered by Google car，the Internet of Vehicles(IoV)will be a distributed transport fabric capable of making its own decisions about driving cus?tomers to their destinations.Like other important instantiations of the Internet of Things(IoT)(e.g.，smart building)，IoV will have communications，storage，intelligence，and learning capa?bilities in order to anticipate customer intentions.The concept that will help in the transition to IoV is vehicular cloud，the equivalent of Internet cloud for vehicles.Vehicular cloud will provide all kinds of the services required by connected vehi?cles.

The urban fleet of vehicles is rapidly evolving.Currently，it is a collection of sensor platforms that provide information to drivers and upload filtered sensor data(e.g.，GPS location，and road conditions)to the cloud(Fig.1).However，it will evolve to become a network of connected vehicles that exchange their sensor inputs with each other in order to optimally perform a well?defined function.This function，in the case of autonomous cars，is prompt delivery of the passengers to destination with maximum safety and comfort and minimum impact on the envi?ronment.In other words，one is witnessing in the vehicle fleet the same evolution from Sensor Web(i.e.，sensors are accessi?ble from the Internet to get their data)to IoTs.(The compo?nents with embedded sensors are networked and make intelli?gent use of these sensors.)

In an intelligent home，the IoT is formed by a myriad of sen?sors and actuators that cover the house internally and external?ly.The IoT can manage utilities in the most economical way，with maximum comfort to residents and almost no human inter?vention.Similarly，in the modern energy grid，the IoT formed by large and small components can manage power loads in a safe，efficient manner with operators playing the role of observ?ers.

?Figure 1. Vehicles are mobile sensors that produce massive amounts of data.

In the vehicular network，like in all the other IoTs，when hu?man control is removed，autonomous vehicles must efficiently cooperate so that traffic flow on roads and highways is smooth. It is predicted that vehicles will behave much better than driv?ers，and more traffic will be handled with fewer delays，less pollution，and more driver and passenger comfort.However，the complexity of distributed control of hundreds of thousands of cars cannot be taken lightly.In the event of a natural disas?ter，vehicles must be able to coordinate to rapidly evacuate crit?ical areas in an orderly manner.This requires efficient commu?nication between vehicles and the ability to discover the loca?tion of needed resources(e.g.，ambulances and police vehicles，information about escape routes，and images about damage that must be avoided).

An efficient communication and distributed processing envi?ronment can be provided by a new network and compute para?digm specifically designed for vehicles:vehicular cloud.This mobile cloud provides several essential services—from rout?ing，content search，spectrum sharing，dissemination and at?tack protection to connected vehicle applications via standard，open interfaces that are shared by all auto manufacturers.

This article discusses the evolution from intelligent vehicle grid to autonomous，Internet?connected vehicles and vehicular cloud.In particular，we highlight the advantages of the Internet of Autonomous Vehicles and discuss related challenges.

2 Characteristics of IoV Applications

Applications in vehicle communications have ranged from safety and comfort to entertainment and commercial services. This section discusses four noticeable characteristics in emerg? ing vehicle applications and offers a vision for intelligent vehi?cle grid and its impact on the autonomous vehicle.Specifical?ly，we base our observations on actual experiments run in Ma?cao，where we operate an urban sensing testbed，and at Univer?sity of California，Los Angeles(UCLA)，where we operate a ve?hicular V2X testbed[1]，[2].We observed the following infor?mation characteristics that are common across a large number of application scenarios.

2.1 Information Characteristics of IoV Applications

Vehicles are data“prosumers”;that is，they have a plethora of sensors that produce a great amount of content，and at the same time，these vehicles consume content from other cars and the Internet.There are several common properties in terms of locality and lifetime.Here，we identify three general properties that are common to many application scenarios:

1)Limited lifetime.Vehicle?oriented content has its own tem?poral scope of validity.This implies that the content must be available during its lifetime.For example，road conges?tion information may be valid for only a few minutes or an accident warning must remain as long as road is not cleared.

2)Spatial validity.Car?generated content has a spatial scope of utility to information consumers.In safety applications，a speed warning near a blind intersection is only valid to vehi?cles approaching to the intersection，say within few hundred meters.In a parking scenario，the information has inherent

local validity because users are generally interested in park?ing close in rather than far away.A similar concept applies to a plethora of other vehicular applications，including ad?vanced navigation systems.

3)Interest locality.This indicates that nearby vehicles repre?sent the bulk of potential content consumers.This concept is further extended so as to distinguish the scope of consum?ers.For instance，all the vehicles in the vicinity want to re?ceive safety messages，while only a fraction of vehicles are interested in commercial advertisements.

Time?space validity of the data implies the scalability of the data collection/storage/processing applications，since old data is discarded.It also implies that the data should be kept on the vehicles rather uploaded to the internet，leading to enormous spectrum savings.This property will be key to the scalability for the Internet of vehicles concept，given the huge amount of data collected by autonomous vehicle sensors[3].

2.2 Information-Centric Networking

Vehicle applications are mainly concerned with content it?self，not its provenance.This memory?less property is charac?teristic of VANETs.To check traffic congestion on the fixed in?ternet，one visits a favorite service site.The site’s URL guaran?tees access to ample，reliable information.In contrast，vehicle applications flood query messages to a local area，not to a spe?cific vehicle，and accept responses regardless of the identity of the content providers.In fact，the response may come from a vehicle in the vicinity that has in turn received such traffic in?formation indirectly from neighboring vehicles.In this case，the vehicle does not care who started the broadcast.This char?acteristic is mainly due to the fact that the sources of informa?tion(vehicles)are mobile and geographically scattered.A typi?cal example is an application that is interested in determining if there are any available roadside parking spaces in a specific area.With IP，a server infrastructure is required to store and record the available parking spots[4];however，with the ICN paradigm，it is possible to directly query the network for avail?able spots[5]，[6].

We expect information?centric networking to play a major role in the management and control of autonomous vehicles. There are two reasons for this:first，the autonomous vehicle will travel at high speeds and short distances from neighbors (on highways)and must have up?to?date information from sur?rounding vehicles of up to several kilometers away in order to maintain a stable course[3].Thus，in the content?centric net?working style，the vehicle periodically sends interest messages to receive position，speed，and direction information from the rest of the fleet.Second，in the case of an accident ahead，the vehicle must alert the driver(who may have their attention else?where)so that they have the option of manually intervening.To prepare the driver to take over，the vehicle retrieves photos and possibly video of the accident scene from the cameras of the vehicles facing the accident.Content?centric networking al? lows access to the best cameras with the needed data，without prior knowledge of the cars that are offering the data.

2.3 Vehicular Sharing of Sensory Data

Emerging vehicle applications consume a huge amount of sensor data in a collaborative manner.That is，multiple sensors installed on vehicles record a myriad of physical phenomena. Vehicle applications collect sensor records，even from neigh?boring vehicles，to produce value?added services.The Car?Speak application[3]，for example，enables a vehicle to access sensors on neighboring vehicles in the same way that the vehi?cle can access its own sensors.The vehicle then runs an auton?omous driving application using the sensor collection without knowing who produced what.Extending the sensorial capabili?ties of connected cars enables situational awareness in an“ahead of time”fashion，thus allowing drivers and autonomous vehicles to identify road hazards.Proactive identification of po?tential road dangers is essential to preventing accidents.For example，if we know that just around a sharp corner pedestri?ans are j?walking，then a potentially deadly accident could be averted.In the general perspective of an intelligent transport system，vehicles exchange traffic congestion and road condi?tion messages to construct an up?to?date traffic and road condi?tion database from which best path to(local)destinations are computed.Collaboration in the sharing and processing of sen?sor data will be one of the strong advantages of autonomous ve?hicles.Continuous sharing of position data is essential to guar?antee the stability of the autonomous fleet.The crowdsourcing of road conditions，such as poor pavements，obstacles，and ac?cidents，using the collection of available sensors will allow smooth driving，even in perilous conditions.Moreover，the col?lective tracking of available channels using sophisticated on board radios will allow careful mapping of the available spec?trum，enabling the efficient communications required for fleet situation awareness and content downloading to“passive”driv?ers.

2.4 Intelligent Vehicle Grid and Vehicular Cloud

Vehicles have sensors that generate copious amounts of data every second.At the same time，the road has smart compo?nents，RFID tags，and embedded microcontrollers[7]，[8]. These“things”constitute a vehicle grid，i.e.，an intelligent road infrastructure analogous to the energy grid for intelligent power generation and distribution.The vehicular grid will have the resources to support services and applications for connect?ed vehicles and autonomous vehicles alike.The various things in the vehicular grid will evolve into the vehicular cloud that provides the computing and communication environment for the Internet of Vehicles.Vehicles become service providers and consumers as well as the infrastructure without an explicit addressing of the resources(i.e.retrieving information from deep?inside the Internet or from a car will be transparent for the user).

The main beneficiaries of vehicular cloud architecture will be autonomous vehicles that drive themselves without human intervention.Such vehicles must be capable of sensing their surroundings and locating routes as well as obstacles.An ad?vanced autonomous car processes all sensory data gathered on?board and from other vehicles[3]，identifies appropriate paths，and constructs a decision tree in order to avoid obstacles on these paths.The vehicular cloud will provide the ideal system environment for the coordinated deployment of the sensor ag?gregation，fusion and database sharing applications required by the IoVs.

3 IoV Connectivity Challenges

There are several challenges to be overcome before the IoV moves from the research prototype stage to commercial deploy?ment.In particular，IoV nodes need to cope with Internet ac?cess，spectrum scarcity，mobility，intermittent connectivity and scalability.In the remainder of this section，we briefly report on recent advances in internet access as well as mobility man?agement achieved through the use of multiple wireless technol?ogies and substituting IP with the named data networking (NDN)information paradigm[9].

3.1 In-Vehicle Connectivity

Several wireless technologies will be required in order to support IoV at scale.The nature of the information(i.e.spatial，temporal，and interest locality)，the requested bandwidth，and channel contention mean that a single solution cannot be re?lied on.Car manufacturers，network operators，and cellular pro?viders will need to provide a plethora of options for the con?nected vehicle[10].

In?vehicle connectivity will be guaranteed by wireless tech?nologies，including IEEE802.11p，Wi?Fi，and 3G/LTE.The networking stack on the vehicle head unit will be in charge of selecting the appropriate communication channel(s)according to current status，user preferences，and application needs.At scale，metro?scale Wi?Fi will be crucial for reducing the load on the cellular infrastructure，especially during rush hours when vehicular density is very high[11]-[13].In the initial phase，we expect IEEE802.11p(DSRC)to play a major role in V2V applications，such as safety and local information retriev?al.Massive deployment of DSRC roadside infrastructure is not expected any time soon because of the massive investment re?quired.In contrast，there are many existing city?scale Wi?Fi deployments.In initial deployment scenarios，high?bandwidth Internet connectivity will be facilitated by cellular and Wi?Fi Infrastructures.Mobile operators are also running out of cellu?lar spectral resources and are encouraging their customers to use the extensive Wi?Fi network wherever possible.This trend is particularly pronounced in France，where all mobile opera?tors offer free Wi?Fi traffic and other incentives to customers who accept to be part of a community of Wi?Fi Hot?Spots1France network operators encourage residential users to offer a slice of their home?con?nection to other fellow customers in exchange of the similar privileges.The authentication is operator controlled.This“community”based approach allowed France network opera?tors to grow their WiFi networks quickly and substantially at no cost..The Wi?Fi protocol，however，was not designed to handle high?speed mobility and frequent connection/disconnections of mov?ing vehicles.The initial connection(attach)time required to ac?cess a Wi?Fi hot?spot is prohibitive for a moving car[14].Even at a moderate urban speed of 35 km/h，the time required to connect，authenticate，acquire an IP，and finally transmit and receive packets on a standard Wi?Fi network is generally too long to be useful[14].The car will likely leave the coverage ar?ea before it has any chance to communicate.An enabler to the IoV vision is the Wi?Fi fast?attach.

An initial approach to the quick access point attach was ex?plored in Quick Wi?Fi[14];however，the proposed solution is limited.It relies on the wireless interface monitor mode that limits the operation to the basic rate(usually 1 Mbit/s).It lacks portability because it uses hardware?dependent primitives，and it does not support any authentication method.Recent research on the fast?attach mechanism done at UCLA and at LIP 6 in Paris has shown promise in terms of resolving the Wi?Fi initial?attach problem.We redesigned the software component that manages Wi?Fi connectivity in Linux/Androd systems，i.e.，the WPA supplicant，in order to achieve quick initial connections and integrate the most common authentication methods.WPA fast is a redesigned WPA supplicant component that uses ad?vanced scanning algorithms that can learn from the environ?ment in order to estimate the amount of time spent channel scanning.Furthermore，it combines the attach process with the authentication process，and this reduces connection time.Al?though detailed description of the WPA?fast protocol is out of the scope of this paper，preliminary experiments performed at UCLA show that WPA?fast cuts the initial attach time needed for a hot?spot connection by one order of magnitude compared to today’s standard WPA.In particular，we drove about two miles across Los Angeles departing from the UCLA campus across a residential area.We tried to connect to any Open or Eduroam authenticated network(Eduroam authentication is based on IEEE802.1x+Radius).WPA Fast was constantly able to connect showing a one order of magnitude improve?ment.In particular，the 80th percentile connection latency for a vehicular Wi?Fi client reduced from 10 s to 0.6 s(Fig.2).

Although WPA?fast is still far from being finalized，it pro?duces encouraging results.Once it has been finalized and ex?tensively tested in both Android and Linux devices，it will con?tribute to removing one of the main roadblocks to scalable，in?expensive high?speed connectivity in connected vehicles.

3.2 Named Data Networking

Scalability issues have traditionally affected mobility over IP.When the IP address changes frequently，such as in com?munity Wi?Fi，mobile IP protocols suffer from high overhead in

both the registration process and routing process.Recently，Van Jacobson et al.[9]proposed a radical change to the Inter?net architecture and shifted the paradigm from a node?centric network to a content?centric network.They observed a huge gain in efficiency by using distributed in?network storage.The NDN architecture proposes to directly address content at the network layer using hierarchical names.The IP layer is re?moved and substituted by the object name.NDN revolves around the pull model in which consumers are first?class citi?zens that request a content piece though a small special packet called interest?packet.For each interest issued by a consumer，the network searches for a matching content unit.If found，the content is delivered to the consumer as an interest response. Routing and forwarding are re?engineered to work on the NDN namespace[9].An initial version of the NDN forwarder was re?leased in 2009，and at present，several state?of?the?art applica?tions have been developed for the NDN architecture，including multiuser video conferencing，web browser，multi?room chat programs，and IoT applications[15]，[16].In a connected car scenario，NDN naturally addresses mobility.When a car moves from an access point to another，it only takes the con?sumer the cost of resending the last not satisfied interest to re?start the communication from where it was left.Network cach?ing prevents wasteful retransmission，and exploiting the broad?cast in the wireless segment enables the potential of the shared medium to be fully exploited.

We designed and developed NDN for vehicular scenarios [6].The protocol was evaluated using a small testbed and simu?lations.The actual experiments were carried at UCLA using ten cars.We identified several application scenarios，including real?time picture transmission and retrieval of real?time traffic information from the road[6].A detailed report on the protocol design and experimental setup can be found in[6]and[17].In a mobility scenario，it was possible to send medium?sized pic?tures(～100KB)using two different wireless connections，a Wi?Fi hot spot，and an ad?hoc connection through a moving relay car.In Fig.3，the color of the dots offers a qualitative hint on how the data was transferred across different interfaces.In con?trast with TCP/IP V?NDN does not need any special mecha?nism to address the client mobility，as it is part of the architec?ture design to the benefit of simplicity and scalability.

The experimental results collected using the UCLA testbed provided a number of insights on the potential advantages pro?vided by Named Data in vehicular environment.In particular，the in?network caching features of V?NDN and the per?packet enable to take full advantage of mobility and to cope with inter?mittent connectivity and disruption.The in?network cache was able to satisfy about one third of the total requests thus reduc?ing the network load and the application latency.Furthermore，in the experiment above was possible to use either one or the other interface or in some cases both of them seamlessly.NDN interest?data mechanism allows to change network interface at with a per?interest granularity.This approach，in contrast with Mobile IP，requires no triangular routing and no additional net?work components.

▲Figure 2.WPA Fast performance in an open network.The chart shows the cumulative distribution function of the time needed to connect to a new access point.

▲Figure 3.Real?time retrieval via V?NDN.

4 Final Remarks

In this paper，we introduced the Internet of Vehicles，which is poised to become a reality with the heavy deployment of con?nected vehicles(as mandated by the US Department of Trans?portation)[18].We identified the main characteristic of the IoV information flow and exposed two of the roadblocks to its deployment，arguing that the ability to exploit the community Wi?Fi connectivity and leave shift from a node?centric para?digm to a content centric paradigm can accelerate deployment and reduce total costs.

[1]G.Pau and R.Tse，“Challenges and opportunities in immersive vehicular sens?ing:lessons from urban deployments，”Image Communication，vol.27，no.8，pp. 900-908，Sept.2012.doi:10.1016/j.image.2012.01.015.

[2]M.Gerla，，J.?T.Weng，E.Giordano，and G.Pau.“Vehicular testbeds?model vali?dation before large scale deployment，”Journal of Communications，vol.7，no.6，pp.451-457，Jun.2012.doi:10.4304/jcm.7.6.451-457.

[3]S.Kumar，L.Shi，S.Gil，N.Ahmed，D.Katabi，and D.Rus，“CarS?peak:a con?tent?centric network for autonomous driving，”in ACM SIGCOMM Computer Com?munication Review，vol.42，no.4，pp.259-270，Oct.2012.doi:10.1145/ 2377677.2377724.

[4]S.Mathur，T.Jin，N.Kasturirangan，J.Chandrasekaran，W.Xue，M.Gruteser，and W.Trappe，“Parknet:drive?by sensing of road?side parking statistics，”in Proc.8th Int.Conf.Mobile Systems,Applications,and Services，San Francisco，CA，USA，Jun.2010，pp.123-136.doi:10.1145/1814433.1814448.

[5]G.Grassi，D.Pesavento，L.Wang，G.Pau，R.Vuyyuru，R.Wakikawa，and L. Zhang，“VANET via Named Data Networking，”inProc.2nd IEEE Workshop on Name Oriented Mobility(NOM),INFOCOM’14，Toronto，Canada.

[6]D.Pesavento，G.Grassi，C.E.Palazzi，and G.Pau，“A naming scheme to repre?sent geographic areas in NDN，”Wireless Days(WD),2013 IFIP，Valencia，Spain，Nov.2013，pp.1-3.doi:10.1109/WD.2013.6686541.

[7]J.M.Kahn，R.H.Katz，and K.S.J.Pister，“Next century challenges:mobile net?working for‘Smart Dust’，”in Proc.5th Annu.ACM/IEEE Inter.Conf.Mobile Computing and Networking，Seattle，WA，USA，Aug.1999，pp.271-278.doi: 10.1145/313451.313558.

[8]E.?K.Lee，S.Yang，S.Y.Oh，and M.Gerla，“RF?GPS:RFID assisted localiza?tion in VANETs，”in IEEE 6th Int.Conf.Mobile Adhoc and Sensor Systems，Ma?cau，China，Oct.2009，pp.621-626.doi:10.1109/MOBHOC.2009.5336946.

[9]V.Jacobson，D.K.Smetters，J.D.Thornton，M.F.Plass，N.H.Briggs，and R.L. Braynard，“Networking named content，”in Proc.5th Int.Conf.Emerging Net?working Experiments and Technologies，Rome，Italy，Dec.2009，pp.1-12.doi: 10.1145/1658939.1658941.

[10]A.Balasubramanian，R.Mahajan，and A.Venkataramani，“Augmenting mobile 3G using WiFi，”in Pro.8th Int.Conf.Mobile Systems,Applications,and Servic?es，San Francisco，CA，USA，Jun.2010，pp.209-222.doi: 10.1145/ 1814433.1814456.

[11]P.Deshpande，X.Hou，and S.R.Das，“Performance comparison of 3G and met?ro?scale WiFi for vehicular network access，”in Proc.10th ACM SIGCOMM Conf.Internet Measurement，Melbourne，Australia，Nov.2010，pp.301-307. doi:10.1145/1879141.1879180.

[12]X.Hou，P.Deshpande，and S.R.Das，“Moving bits from 3G to metro?scale WiFi for vehicular network access:an integrated transport layer solution，”in 19th IEEE Int.Conf.Network Protocols，Vancouver，Canada，Oct.2011，pp. 353-362.doi:10.1109/ICNP.2011.6089074.

[13]U.S.Department of Transportation.(2014，Jan.).Vehicle?to?Infrastructure(V2I) Communications for Safety[Online].Available:http://www.its.dot.gov/research/ v2i.htm

[14]J.Eriksson，H.Balakrishnan，and S.Madden，“Cabernet:vehicular content de?livery using WiFi，”in Proc.14th ACM Int.Conf.Mobile Computing and Net?working，San Francisco，CA，USA，Sept.2008，pp.199-210.doi:10.1145/ 1409944.1409968.

[15]W.Shang，J.Thompson，M.Cherkaoui，J.Burke，and L.Zhang，“NDN.JS:a Ja?vaScript client library for named data networking，”in Proc.IEEE INFOCOMM 2013 NOMEN Workshop，Turin，Italy，Apr.2013.

[16]D.Kulinski and J.Burke，“NDN video:live and prerecorded streaming over NDN，”NDN Project Team，Tec.Rep.，2012.

[17]G.Grassi，D.Pesavento，L.Wang，G.Pau，R.Vuyyuru，R.Wakikawa，and L. Zhang，(2013，Oct.22).Vehicular Inter?Networking via Named Data[Online]. Available:http://arxiv.org/pdf/1310.5980.pdf

[18]NHTSA.(2014，Feb.3).U.S.Department of Transportation Announces Decision to Move Forward with Vehicle?to?Vehicle Communication Technology for Light Vehicles[Online].Available:http://www.nhtsa.gov/About+NHTSA/Press+Re?leases/2014/USDOT+to+Move+Forward+with+Vehicle?to?Vehicle+Communi?cation+Technology+for+Light+Vehicles

Manuscript received:February 25，2014

Biograpphhiieess

Mario Gerla(gerla@cs.ucla.edu)obtained his undergraduate engineering degree from Politecnico di Milano，Italy.He received his PhD degree from UCLA.In 2002，he became a fellow of the IEEE.As a graduate student at UCLA，he was part of the team that worked on the early ARPA Network system and protocols under the guid?ance of Professor Leonard Kleinrock.After four years at Network Analysis Corpora?tion，New York，he joined UCLA in 1976.At UCLA he designed network protocols，including ad hoc wireless clustering，multicast(ODMRP and CODECast)and Inter?net transport(TCP Westwood).He has lead the ONR Mimuteman project，designing the next?generation scalable airborne Internet for tactical and homeland defense sce?narios.He is now leading several advanced wireless network projects funded by in?dustry and government.His team is developing a vehicular testbed for safe naviga?tion，content distribution，urban sensing and intelligent transport.Parallel research activities are wireless medical monitoring using smart phones and cognitive radios in urban environments.He has served on several conference program committees，including MobiCom，MobiHoc，MedHocNet and WONS.He is on the IEEE TON sci?entific advisory board.

Giovanni Pau(giovanni.pau@lip6.fr)is the ATOS/Renault smart mobility chair pro?fessor at the University Pierre at Marie Curie，Paris.He received the Italian Laura in computer science in 1998.He received his PhD degree in computer engineering from the University of Bologna in 2002.Before Joining UPMC，Dr.Pau was a senior research scientist at the UCLA Computer Science Department，where he is currently an adjunct professor.

Dr.Pau’s core research interests include network systems with a focus on vehicular networks and pervasive mobile sensor systems.He designed and built the UCLA campus vehicular testbed and the UCLA/MPI urban sensing testbed designed to en?able hands?on studies on vehicular communications and urban sensing.His re?search contributions lead to the VERGILIUS and CORNER simulation suites de?signed to support mobility and propagation modeling in urban environments.More recently，Dr.Pau designed and developed VNDN the Named Data Network(NDN) protocol stack specifically adapted to work on mobile?to?mobile scenarios.

Rita Tse(ritatse@ipm.edu.mo)is the program coordinator in the Computing Pro?gram at the Macao Polytechnic Institute.Her primary research interests are on ubiq?uitous computing and urban sensing.She published several papers in esteemed in?ternational journals and conferences.She is currently serving as local manager for the UCLA?MPI Joint Research Laboratory in Ubiquitous Computing.

ATOS and RENAULT through the UPMC Chair on Smart Mobility,the MPI Research Fund,and the National Science Foundation through the GREEN?CITY project support this work.