Device-to-device Communication

Transcription

1 COMP9336/4336 Mobile Data Networking or ~cs4336 Device-to-device Communication 1

2 Lecture overview There is a growing need and motivation for devices to communicate with other nearby devices, directly. This lecture focuses on the communication solutions available for direct device-to-device communication. 2

3 Reliance on base station Today s mobile communication relies on base stations Base station/access point/infrastructure One device communicates with another via BS Pros: Simple device functionality, greater/central control of the network, long distance reach Cons: Devices within direct wireless range of each other still need to involve the BS for local communications BSs are getting overloaded as device numbers explode Less efficient use of spectral resources ( local use of spectrum not possible) 3

4 D2D Communication D2D means two devices within wireless range can communicate directly with each other without any help from a BS Benefits include: Offloading local communications from the overloaded BSs Potential for more efficient use of spectral resources Lower latency Both WiFi and Cellular, two most popular communication infrastructures, can benefit from D2D 4

5 Lecture contents Unlicensed D2D NFC (near field communication) Bluetooth WiFi Direct Licensed D2D LTE Direct DSRC (dedicated short range communication), also known as vehicle-to-vehicle (V2V) communication 5

6 NFC 6

7 NFC Recently introduced as a pure/native D2D technology Very short range only a few cm (close to contact) MHz (same as RFID frequencies) Unlicensed use of spectrum Very low power (15mA) Very low data rates Kbps Transfers small amount of data quickly (no pairing needed) Applications Mobile payment Ticketing Business card exchange Smart posters (passive NFC/RFID tags implanted in the poster) And so on 7

8 Bluetooth 8

9 Bluetooth Characteristics Operates (coexists) in the same unlicensed band (2.4GHz) as wifi Designed exclusively for D2D (A pure, native, and oldest D2D technology) Numerous successful applications, including headset and car speaker for mobile phones Low power consumption (less than 100 mw) and low data rates (around 1Mbps) Suitable for energy-constraint devices and for applications not requiring a huge bandwidth Very short range: 10 meters Medium access control is centralised (one node, called Master, controls several slave nodes) 9

10 Bluetooth channels 79 channels each 1-MHz wide (2402 MHz to 2480 MHz) Three scanning (advertising or enquiring) channels (37-39) Figure source: 10

11 Frequency Hopping (1) Large number of devices need to share the same ISM frequencies (interference becomes and issue) Bluetooth uses a concept called Frequency Hopping switch the frequency for every packet transmission (every slot) WiFi is a non-hopping ISM system (continues to transmit packets after packets over the same frequency unless the AP is reset with a new channel) Ideally could switch to any of the 79 frequencies, but sometimes the hopping is restricted to a smaller subset (such as 23 frequencies) 11

12 Hopping pattern (hopping sequence) The transmitter and the receiver must operate in the same frequency to receive a packet successfully Therefore, both the Tx and the Rx must know when to switch to which frequency The transmitter (master) generates a pseudorandom frequency hopping sequence and share this with the receiver at the start of the session to ensure that both Tx and Rx remain synchronised at all times 12

13 Piconet Transmission is controlled centrally by a master node Upto 7 slave nodes may join a master node to form a piconet Each slave is identified by a unique address Salves (receivers) listen to every slot because they do not know which slot will be used by the master (transmitter) Master (transmitter) transmits only when it has data to transmit Transmitter therefore usually consumes less power than receiver in a given communication scenario (asymmetric power consumption) master slave 13

14 Piconet formation All nodes are scanning for inquiries A node interested in communicating with other devices sends a query (becomes a master) Any node that responds becomes a slave Master establishes connection to each slave one by one Up to 7 nodes can join a piconet Slaves synchronise themselves with the frequency hopping pattern of the master 14

15 Scatternet A node can act as both master for one piconet and slave for another (multiple piconets overlap) Piconet 1 master master/ slave Piconet 2 slave 15

16 WIFI DIRECT 16

17 Motivation for WiFi Direct Bluetooth is a great success story for D2D communication, but it has two major issues pairing is cumbersome Date rate is too slow for some large data transfer applications (photo sharing, video games, ) Pairing problem is addressed by NFC, but with further reduction in data rates WiFi Direct solves the data rate issue WiFi direct is basically wifi, but it allows two devices in proximity to exchange data without having to connect to infrastructure (i.e., AP as a third device) 17

18 How does wifi direct work? At least one device needs to be wifi direct capable Wifi direct capable means it can act as an AP (an embedded software in the wifi radio) The other device then can connect to the device capable of acting as an AP Wifi direct uses WiFi Protected Setup (WPS) 18

19 WiFi Protected Setup (WPS) Usual AP setup requires manual interventions Choosing an SSID Choosing a password phrase Manual setup is a barrier to instantaneous and short-lived D2D sessions, such as transferring a video clip WPS makes these manual procedures automatic Automatic selection of SSID Automatic selection of security keys 19

20 LTE Direct 20

21 Motivations for LTE Direct NFC/Bluetooth/WiFi provide short range D2D (10cm m) A longer range and more ubiquitous D2D would enable new applications Cellular networks are getting overloaded D2D is seen as a potential technology to help offload traffic from overloaded cellular networks LTE, the latest standard in cellular, is now considering a D2D option called LTE Direct 21

22 LTE Direct Architecture D2I device to infrastructure D2D device-to-device Both D2I and D2D share the same spectrum Frequency allocation for both D2I and D2D is under the control of the network BS D2I Local Communication 22

23 LTE Direct interference LTE Direct uses the same licensed band as D2I communication (i.e., the LTE band) Interference therefore is a major issue to be managed D2D resource allocation is therefore managed by the network, which has detailed knowledge of the locations of the devices and the use of other frequencies in the vicinity 23

24 DSRC (V2V) 24

25 How deadly is Road Crash? One of the leading causes of death 1.2 million dies every year 9 th in 1990, could become 3 rd in nd after HIV/AIDS for young men 50 million people injured every year In US, road accidents cost $250 billion/year Insurance, ambulance, hospital, etc. 25

26 Is road crash preventable? Most of the collisions could be avoided if the driver was provided early warning prior to the collision Question: Can we develop an effective and affordable crash-warning technology for next generation vehicles? Answer: Perhaps (if it was possible for vehicles to communicate with each other in real-time) 26

27 Motivation for V2V (1) current safety systems have limitations Today s safety systems are designed to survive crashes Seat belt, airbag, side-impact bars, anti-lock brakes, We need survival as well as prevention Famous idiom prevention is better than cure Crash prevention would require crash-warning systems Independent (non-communicative and non-cooperative) sensorbased (radars, lidars, vision) crash warning systems are complex, expensive, and unreliable Complex signal processing for motion detection and tracking Limited field-of-view, cannot penetrate obstacles (difficult to track over-the-hill, around-the-corner, behind-a-big-truck) Very challenging to track multiple object simultaneously High percentage of false alarms Cannot be used for any other purposes (no cost amortization) High cost limits their use to exclusive luxury vehicles 27

28 Motivation for V2V (2) cooperative & beyond line-of-sight driving It becomes a lot easier to reliably predict crashes if motion data (location, velocity, acceleration, direction, etc.) is available Motion equations to compute future locations of moving objects Large number of objects can be tracked simultaneously Motion data of nearby vehicles can be obtained by direct V2V communication Unlike independent radars, vehicular communication enables cooperative crash prevention Vehicles cooperate with each other to prevent collisions Using wireless radio, we can overcome around-the-corner, over-thehill, and behind-a-big-truck problems Better awareness of the surrounding 28

29 Motivation for V2V (3) beyond safety V2V can support a wealth of other (non-safety) applications helps with cost amortization and affordability V2V will address critical socio-economic issues by Reducing travel time, and fuel consumption (climate change) through cooperative driving V2V can be used for electronic payments (e-toll etc.) Vehicle-to-infrastructure communications V2V can be leveraged for commercial services Digital map download, Internet access for the vehicle, 29

30 Achievements and Developments (1) The vision of cooperative driving had two caveats (1) Wireless spectrum, and (2) Car interoperability Development in wireless spectrum Release of 75MHz spectrum (in 5.9GHz) by US FCC Achievement in interoperability Formation of automaker consortia Involvement of standard bodies (IEEE, ) 30

31 Achievements and Developments (2) Gaining irreversible momentum engagement of all stake holders (govt., automaker, standard body) is a significant step forward VC is not a question of if but when Decision on base technology reusing IEEE /WiFi/WLAN) Release of standard drafts (e.g p) Completion of preliminary field trials confirmation of technical viability 31

32 Global Initiatives (1) Countries leading the efforts US, EU (Germany, ), Japan (the largest automakers of the world) Sectors involved Government (US Dept. of Transportation, ) Automakers (Toyota, BMW, Mercedes, Daimler Crysler, GM ) Standard bodies (IEEE, ) Academia 32

33 Global Initiatives (2) US Vehicle Infrastructure Integration: Crash Avoidance Metrics Partnership: www-nrd.nhta.dot.gov/ departments/nrd-12 EU Communication for esafety ( Secure vehicle communication ( Japan Advanced safety vehicle project ( 33

34 System Architecture V2V-based Crash Avoidance 34

35 Communication-based crash avoidance the basic architecture wireless communication (motion data Tx/Rx) Driver crash warning OBU Vehicle Electronics (motion data and control) 35

36 The operation of V2V Based Crash Avoidance Vehicles are equipped with onboard units (OBUs) interfaced with vehicle electronics (access to motion data and control) short-range communications unit (exchange motion data with other cars) positioning capability, e.g., GPS (to determine its current position) onboard computer (motion data analysis and crash prediction) Using the communication unit, vehicles frequently (~10Hz) exchange location and motion data with other nearby (~100m) vehicles achieves 360 degree awareness Onboard computer continuously analyses the motion data received from nearby vehicles warns the driver if it detects a collision in progress Driver (and/or the car) takes actions to prevent crash or mitigate the after effect of crash automated car actions: pretension the seat belt, apply brake, 36

37 Crash avoidance scenario (1) slow vehicle warning 37

38 Crash avoidance scenario (2) sudden halt warning 38

39 Crash avoidance scenario (3) intersection warning A approaches intersection on Green at 100 Kmh B, possibly drunk, about to enter intersection on RED A & B would crash with current technology Too late for A to stop when she sees B right in front of her With vehicular communication, A is warned in time about possible RED light violation by B A approaches intersection more cautiously, manages to stop in time, and avoids the crash 39

40 Applications of V2V Communication 40

41 Safety Applications Crash warnings on entering intersections, railway crossings entering and departing highways sudden halt lane change 41

42 Non-safety Applications Other than safety, vehicular communication could also Improve driving efficiency, e.g. notification of accidents up the road, real-time congestion information, Enhance driver convenience and comfort, e.g. toll & parking payment Assist drivers, e.g. parking in a difficult spot, advanced cruise control, lane keeping, roadsign recognition, Create new commercial opportunities, e.g., road-side business advertisement (gas station, fast food), content downloading (digital map, music), Internet access, Automate highways ( autopilot ) Many non-safety applications would require vehicle-toinfrastructure (V2I) communications 42

43 Requirements of V2V Communication 43

44 Requirements of V2V Communications safety High mobility ( Kmh) Low bandwidth (100Byte packet, 10 packet/second) Low latency (Delayed information is useless) High reliability (sensitive to packet loss) One-to-all (broadcast) communication 44

45 Requirements of V2V Communications non-safety High mobility High bandwidth (content downloading) One-to-one (unicast) communication Less sensitive to packet loss similar to Internet access Short connection/association establishment time 45

46 Meeting the requirements technology decision Developing a brand new technology is costly, and its performance would be largely untested in real environment Would be worth it if existing technologies could be reused 46

47 Can Cellular (3G) meet the requirements? Communication range: thousands of meters most roads are covered by cellular radio towers Infrastructure mode only communications can only happen between mobile terminals and base stations (mobile-to-mobile not supported) To broadcast, the terminal sends to base station and the base station sends to other terminals End-to-end latency: seconds The broadcast and latency requirement exclude cellular technology for vehicular safety communications (cellular could be used for non-safety) 47

48 Can meet the requirements? Communication range: ~300m Support both infrastructure and ad-hoc mode Easy to broadcast to many nearby receivers Latency: 10 2 ~10 3 micro seconds is more promising for the vehicular context (As we ll see shortly, DSRC is basically an extension of optimized for vehicles) 48

49 Technical Standard for V2V Communications DSRC Direct Short Range Communication 49

50 IEEE Standards (DSRC) communication stack DSRC refers to a communication stack optimized for vehicular environment IEEE 1609 (WAVE) IEEE p (DSRC) PHY and MAC Layers IEEE p IEEE Networking, Transport, Presentation, Session Layers IEEE PHY+MAC DSRC spectrum Wave wireles access for vehicular environment 50

51 DSRC Spectrum and Channels The US FCC has allocated 75MHz spectrum at 5.9GHz for DSRC The spectrum is divided into 7 10MHz channels (Channel 172 ~184) Channel 174&176, and channel 180&182 can be combined to form 20MHz channels Channel 178 is the control channel (CCH) reserved for safety communications Non-safety communications use the other 6 service channels (SCHs) 51

52 DSRC Spectrum Allocation Canadian Special License Zones* US Spread Spectrum Allocation Control Channel Service Channels Reserved for harmonization with potential extension of the UNII band US and Potential Mexican DSRC Allocation Optional 20 MHz # Optional 20 MHz # Ch 172 Ch 174 Ch 176 Ch 178 Ch 180 Ch 182 Ch 184 Proposed Canadian DSRC Allocation Optional 20 MHz # Optional 20 MHz # Potential UNII Expansion Channels Ch 172 Ch 174 Ch 176 Ch 178 Ch 180 Ch 182 Ch Frequency (GHz) 2013 * The Mahbub use of Hassan, channels UNSW overlapping these zones may be restricted in some locations in Canada. # 10 MHz Channels with 20 MHz combination options 52

53 PHY and MAC Layer PHY/MAC functionalities are defined by p and IEEE p is based on a with several variances p Physical layer Main difference: 20MHz channel (11a) 10MHz channel (11p) Halved channel bandwidth» Halved data rate (max. 27Mbps)» reducing the Doppler spread caused by high mobility» doubling the symbol duration: multipath interference reduced Mac Layer supports priority (similar to e) reduces AP association overhead (unicast service channel communication) has a simplified ad hoc mode: no association needed (for safety broadcast on control channel) [note that WiFi Direct needs association] Dynamic MAC address: preventing tracking vehicles (privacy)» Generating a random MAC address when device starting up» Generating a new MAC address when address conflicting detected 53

54 Supporting multichannel operation (channel switching in DSRC) was designed for single channel operation all clients tune to the channel used by the AP all communication use the same channel DSRC uses multiple (all 7) channels concurrently Problem: a radio cannot monitor two channels at the same time One option is to manufacture vehicles with multiple dedicated DSRC radios for each (categories of) channel For vehicles with single radio communication systems, channel switching is required to concurrently support both safety (control channel) and non-safety (service channels) applications over the same DSRC unit IEEE specifies how switching between channels is done 54

55 DSRC Channel Switching with IEEE All stations are time synchronized using GPS All stations switch to and leave CCH at the same time stay in CCH for CCH_interval before switching back to service channels stay in SCH for SCH_interval before switching back to control channel Guard Interval: accounts for channel switching delay and synchronization variance Sync Interval should be short (tens of milliseconds) to allow vehicles exchange motion data frequently Exact values of the intervals are yet to be specified 55

56 WAVE Network/Transport Layer IEEE TCP/IP only for service channels mainly for non-safety unicast communication For single-hop motion data broadcast, TCP/ IP header information is useless (no routing, flow control, etc. needed) unnecessary latency and channel overhead WAVE Short Message Protocol (WSMP) alternative to IP for safety communication optimized for safety broadcast on CCH applications can directly control physical layer parameters (channel, power, data rate) non-safety can also use it for short control messages (eg. Service association) on SCHs Non-Safety Safety TCP/UDP WSMP IPv6 WAVE MAC 56

57 Security services IEEE Some services may require secured communication specifies secure message formats and processing Scenarios for secure message exchange 57

58 IEEE (optional layer) Defines data storage formats (among other things) Implementation is optional Similar to OSI presentation/session layer which is hardly implemented in practical systems 58