Future radio access implementation & demonstration Scandinavian workshop on testbed-based wireless research November 27 th 2013

Transcription

1 Future radio access implementation & demonstration Scandinavian workshop on testbed-based wireless research November 27 th Wireless Access Networks, Ericsson Research, Stockholm

2 goal To study and investigate new wireless communications mechanisms via prototyping and proof-of-concept FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 2

3 content Platform: the WARP board Hardware: pcores Software: elements Real-world Concepts to-date Timeline Demonstration Conclusion FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 3

4 Platform: warp board Off-the-shelf wireless prototyping kit RICE support: reference design Online community support Ethernet interface, ISM-band radio Dumb electronics around a smart FPGA (Xilinx Virtex II Pro) FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 4

5 Platform: warp board Off-the-shelf wireless prototyping kit RICE support: reference design Online community support Ethernet interface, ISM-band radio Dumb electronics around a smart FPGA (Xilinx Virtex II Pro) FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 5

6 Platform: warp board Off-the-shelf wireless prototyping kit RICE support: reference design Online community support Ethernet interface, ISM-band radio Ctrl Data CPU IC DMAC RAM Fixed Dumb electronics around a smart FPGA (Xilinx Virtex II Pro) FPGA: fixed portion: PC-like Flexible flexible portion: PC s peripherals (pcores) FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 6

7 Platform: warp board Off-the-shelf wireless prototyping kit RICE support: reference design Online community support Ethernet interface, ISM-band radio Dumb electronics around a smart FPGA (Xilinx Virtex II Pro) FPGA: fixed portion: PC-like flexible portion: PC s peripherals (pcores) Build pcores with the flexible, write C code to run on the fixed to interact with pcores Ctrl Data CPU SISO OFDM IC DMAC RAM 100BaseTX FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 7

8 Hardware: PCORE SISO OFDM Home-grown: developed together with our university UCSD Single OFDM radio, 4MHz bandwidth, ISM band operation (MAX2829 chip) Ctrl Data CPU IC DMAC RAM 100BaseTX SISO OFDM FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 8

9 Hardware: PCORE SISO OFDM Home-grown: developed together with our university UCSD Single OFDM radio, 4MHz bandwidth, ISM band operation (MAX2829 chip) Ctrl Data CPU IC DMAC RAM 100BaseTX SISO OFDM FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 9

10 hardware: PCORE SISO OFDM Home-grown: developed together with our university UCSD Single OFDM radio, 4MHz bandwidth, ISM band operation (MAX2829 chip) Ctrl Data CPU IC DMAC RAM 100BaseTX SISO OFDM FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 10

11 Software: elements Home-grown: mainly written in Kista (embedded programming, C) Low-level protocols: Radio MAC: addressing and power-control IP/ethernet for wired communications Layer-2 bridge: SISO OFDM ethernet Control modules for interacting with running board, changing configuration on-the-fly Same software on all nodes (differentiation via DIP switch) Support modules for data export (i.e., for logging, visualization, etc.) Kernel Radio MAC SISO OFDM User control Concept Data Export Packet filter Ethernet FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 11

12 Real-world: nodes WARP board = Layer-2 ethernet bridge: Wireless SISO OFDM ç è wired 100BaseTX Tether WARP-board to a FreeBSD PC via dedicated 100BaseTX ethernet PC: higher layers WARP board: lower layers Wireless node = PC + WARP board Ethernet 100Mbps UTP Network (IP) Routing over WAN Radio MAC Addressing over-the-air PHY (SISO) QPSK, 1/3-coding, sync, etc. Over-the-air wireless node-to-node IP communication possible! Wireless Node Wireless Node FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 12

13 real-world: the lab Lab Part of Ericsson Research, Wireless Access Networks Located in Kista, Stockholm FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 13

14 Concepts: to-date Concepts demonstrated to-date: Autonomous relay mechanism In-between node auto-relays when SRC-DST endpoint communications fail Uplink CoMP UE uplink transmission detected by 2 BS I/Q samples exchanged via BSs ethernet backhaul and MRCcombined for better performance RSSI sampler for Wi-Fi load studies Estimate Wi-Fi 2.4GHz channel utilization based on RSSI values Bi-directional network coding Measurements of over-the-air XOR network coding behaviour Network-assisted device-to-device communications FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 14

15 Concepts: d2d Applies to devices in close proximity Exploit radio s natural broadcast characteristic Single-hop vs. multi-hop via the network Shorter distancesè lower transmit power Higher modulation/bitrates Offload network traffic General goodness overall! 100m 100m 1m Classic-mode 1m D2D-mode FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 15

16 Concepts: na-d2d Networked society: 50 billion diverse devices Alarm clock awakens coffee maker Washing machine notifies tumble dryer to warm up Advantages of network-controlled Network is ubiquitous: works everywhere Zero configuration - no need for SSIDs, WEP/WPA, discoverable, passkeys, etc. Licensed spectrum operation DSLR TV example: sequence of events Communications commence in classic cellular manner Network discovers endpoints are in close proximity Endpoints are reconfigured, resources re-assigned Benefits of D2D enjoyed! FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 16

17 Concepts: na-d2d Networked society: 50 billion diverse devices Alarm clock awakens coffee maker Washing machine notifies tumble dryer to warm up Advantages of network-controlled Network is ubiquitous: works everywhere Zero configuration - no need for SSIDs, WEP/WPA, discoverable, passkeys, etc. Licensed spectrum operation DSLR TV example: sequence of events Communications commence in classic cellular manner Network discovers endpoints are in close proximity Endpoints are reconfigured, resources re-assigned Benefits of D2D enjoyed! FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 17

18 timeline TDMA Cellular Single-cell 2 devices NA-D2D Transit from TDMA cellular to D2D System timeslots savings Distributed power control D2D pair self-manage power control Exploit closer proximity Intermediate relay other node helps out D2D pair Further system-wide power-savings 2010 TEXT TEXT 2011 TEXT 2012 TEXT 2013 TEXT FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 18

19 2010: Tdma cellular Introduction of an over-the-air synchronization overlay Setup consists of a single Master node, and multiple Slave nodes Master node generates unique SYNC beacon every 10ms Slaves listen for SYNC. Upon sync, both have a concept of a TDMA frame è can subsequently determine timeslots Radios are only powered-on for active timeslots (otherwise idle) Software used to configure timeslots Radio power Radio power Slot ID System resource (timeslot) Powered on Master node transmit power for SYNC 10ms Slave node receive power for SYNC Powered Sync d on 10ms TDMA frame: Frame 0 Frame 1 Frame ::: FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 19

20 2010: Tdma cellular We emulate a single-cell TDMA cellular mobile system Three nodes: a BLUE network-node (BS), a GREEN device (TV), and a RED device (camera) The Sequence is as follows Slot ID System resource (timeslot) FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 20

21 2010: Tdma cellular We emulate a single-cell TDMA cellular mobile system Three nodes: a BLUE network-node (BS), a GREEN device (TV), and a RED device (camera) The Sequence is as follows The BLUE BS is 1 st powered on periodic SYNC transmitted every 10ms SLOT0 and SLOT5 allocated for DL and UL control (DLC & ULC) Slot ID System resource (timeslot) FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 21

22 2010: Tdma cellular We emulate a single-cell TDMA cellular mobile system Three nodes: a BLUE network-node (BS), a GREEN device (TV), and a RED device (camera) The Sequence is as follows The GREEN TV is powered on Detects SYNC; understands framing Sends ATTACH-REQ to BS on the ULC BS authenticates TV, then assigns UL&DL data slots via the DLC Slot ID System resource (timeslot) FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 22

23 2010: Tdma cellular We emulate a single-cell TDMA cellular mobile system Three nodes: a BLUE network-node (BS), a GREEN device (TV), and a RED device (camera) The Sequence is as follows Similarly, the RED Camera is powered on Detects SYNC; understands framing Sends ATTACH-REQ to BS on the ULC BS authenticates Camera and assigns UL&DL data slots via DLC Slot ID System resource (timeslot) FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 23

24 2010: Tdma cellular We emulate a single-cell TDMA cellular mobile system Three nodes: a BLUE network-node (BS), a GREEN device (TV), and a RED device (camera) The Sequence is as follows User data generated by GREEN arrives at BS From packet-headers, BS forwards data onward to RED Similarly for data in the RED è GREEN direction Slot ID System resource (timeslot) FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 24

25 2011: Na d2d How to determine direct GREENè RED link? RED receives data on DL slot (SLOT3) Additionally, BS instructs RED via DLC, to also listen to SLOT7 (i.e., eavesdrop on GREEN s UL transmissions) RED should buffer these packets overheard on SLOT7 Simultaneously, BS also buffers packets received on SLOT7 At some point in the future, buffered packets from RED are sent to the BS for the network to evaluate Slot ID System resource (timeslot) FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 25

26 2011: Na d2d How to determine direct GREENè RED link? RED receives data on DL slot (SLOT3) Additionally, BS instructs RED via DLC, to also listen to SLOT7 (i.e., eavesdrop on GREEN s UL transmissions) RED should buffer these packets overheard on SLOT7 Simultaneously, BS also buffers packets received on SLOT7 At some point in the future, buffered packets from RED are sent to the BS for the network to evaluate Similarly, repeat for the REDè GREEN link Slot ID System resource (timeslot) FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 26

27 2011: Na d2d For the direct GREENè RED link, the network compares its own (i.e., BS) measurement, with received RED s report BS s own measurement RED s measurement Wireless node BLUE s console output [ ] D2D slot 7 direct-mode measurements:! [ ] =>Rx SN: ! [ ] =>RSSI : ! [ ] D2D_CTRLMSG_RPT_MEAS received from Node FID=24! [ ] D2D slot 7 direct-mode measurements:! [ ] =>Rx SN: ! [ ] =>RSSI : ! Identical SNs in both measurement, and stronger RSSI in RED s report direct GREENè RED link GOOD! Slot ID System resource (timeslot) FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 27

28 2011: Na d2d For the direct GREENè RED link, the network compares its own (i.e., BS) measurement, with received RED s report BS s own measurement RED s measurement Wireless node BLUE s console output [ ] D2D slot 7 direct-mode measurements:! [ ] =>Rx SN: ! [ ] =>RSSI : ! [ ] D2D_CTRLMSG_RPT_MEAS received from Node FID=24! [ ] D2D slot 7 direct-mode measurements:! [ ] =>Rx SN: ! [ ] =>RSSI : ! Identical SNs in both measurement, and stronger RSSI in RED s report direct GREENè RED link GOOD! Reassign slots, notify device RED via DLC SLOT3 now FREE! Slot ID System resource (timeslot) FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 28

29 2011: Na d2d For the direct GREENè RED link, the network compares its own (i.e., BS) measurement, with received RED s report BS s own measurement RED s measurement Wireless node BLUE s console output [ ] D2D slot 7 direct-mode measurements:! [ ] =>Rx SN: ! [ ] =>RSSI : ! [ ] D2D_CTRLMSG_RPT_MEAS received from Node FID=24! [ ] D2D slot 7 direct-mode measurements:! [ ] =>Rx SN: ! [ ] =>RSSI : ! Identical SNs in both measurement, and stronger RSSI in RED s report direct GREENè RED link GOOD! Reassign slots, notify device RED via DLC SLOT3 now FREE! Similarly for the REDè GREEN direction SLOT2 now FREE! Slot ID System resource (timeslot) FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 29

30 2012: Power control Concept Inter-device distance closer - should require less transmit-power! Challenge Network-independent device-managed power-control mechanism Implemented: MAC-based power-control Incorporated power-control into our simple radio MAC headers Receiver always ACKs received packet Transmitter up/down power - based on ACK/no-ACK MAC header contains RSSI delta of last received packet: other-end always knows how loudly transmitted packet was received FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 30

31 2013: D2d repeaters Concept Thriftier to take smaller hops P SD (P SR + P RD ) << P SD! S D Positive: system-wide power utilization will reduce d d P SR P RD Negative: additional delays, i.e., 2 transmissions from S to D, instead of 1. S R D FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 31

32 2013: D2d repeaters Challenge I How do determine the best R to repeat? Network holds election : Broadcast instruction to all nodes in vicinity, i.e., S, D, R 6, R 7, R 8, R 9! Instruction: eavesdrop on ongoing S D transmissions for duration t eaves! Report back to network: RSSI_AVG S and RSSI_AVG D over period t eaves! The R-nodes will report both RSSI_AVG D and RSSI_AVG S while S and D nodes only report RSSI_AVG D and RSSI_AVG S respectively (see table) Evaluating this the network can determine which R 8 hears BOTH S and D best AND better than S and D hear each other, i.e., the R x in between S and D! S R 6 Network R 8 Node! RSSI_AVG S! RSSI_AVG D! S! -! 200! D! 205! -! R 6! 400! 156! R 7! 150! 300! R 8! 300! 303! R 9! 210! 450! R 9 Measurement reports at the network Values 205 and 200 set the threshold which the R-nodes must exceed to qualify as a repeater. R 8 and R 9 both qualify, with R 8 being a better-fit R 7 D FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 32

33 2013: D2d repeaters Challenge II How to change topology? Exploit MAC-based endpoint controlled power-control mechanism: 1 Network instructs R 8 to be repeater 2 Initially operate at high transmit power to drown out original transmitters, i.e., end- S Data R 8 ACK D points S & D only hear R 8 Since R 8 always responds to endpoints S & D, their power-control mechanism lowers their transmit power This causes D to no longer hear S directly, and vice-versa R 8 then now starts lowering its transmit power Repeated packets are also tagged so end-points are aware repeater is present! S Data ACK R 8 Data ACK D FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 33

34 2013: D2d repeaters Challenge III Mobility and topology changes How to detect and trigger fallback: If topology changes, power-control adaptation will result in D receiving 2 identical packets: 1 from S and 1 from R 8 R 8 will see 2 identical ACKs from D or ACK to a packet which R 8 hasn t repeated yet R 8 should then notify the Network that it s stopping, and the S D communication takes place directly again Does this work?! S Data ACK R 8 Data ACK D R 8 FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 34

35 Demonstration today D2D power control MAC-based Between direct-communicating device-pairs! P SD D2D repeaters D d d S Device-based repeater for multi-hop direct-communicating device pairs P RD P SR D R S FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 35

36 conclusion To study and investigate new wireless communications mechanisms via prototyping and proof-of-concept Designed and prototyped various NA-D2D mechanisms Implementation behaved mostly as expected Discovered subtle real-world issues from implementation (timing, asymmetries) Fed discoveries back to design process more robust mechanism FRA Implementation & Development (FRAID) Updated: 27 th November, 2013 Page 36

37