k J. Schwoerer (France Telecom) – N. Dejean (Elster)) Slide 1 Project: IEEE P Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [Elster & France Telecom proposal] Date Submitted: [15 September, 2011] Source: [Jean Schwoerer, Nicolas Dejean] Company [France Telecom R&D, Elster] Address [28 chemin du vieux chênes FRANCE ] Voice:[ ], FAX: [ ], Re: [.] Abstract:[This document give preliminary information on the proposal that we submit] Purpose:[Description of what the author wants P to do with the information in the document] Notice:This document has been prepared to assist the IEEE P It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein. Release:The contributor acknowledges and accepts that this contribution becomes the property of IEEE and may be made publicly available by P

k Slide 2 Orange - France Telecom / ELSTER Technical Proposal

k Agenda -LECIM devices needs : -Long range -Low power -Proposed network capabilities -Proposed PHY features -RF characteristic -FEC and interleaving -MAC -Coexistence -Conclusion Slide 3

k Slide 4 Sub-GHz wireless connectivity platform Battery powered LECIM devices will provides : –Long range (high link budget) for cost effective network infrastructure and long operating range –Ultra-low power management to reach multi-year operation –Ability to peacefully coexist with other devices –permanent reachability with human acceptable turn-around time (a few 10’s of sec) Large scale wireless sensor network needs : Typical network structure (star and tree) at reasonable cost Mesh / relaying for hard to reach endpoints or to recover wireless connectivity after major events Efficient power management to save endpoint battery life “almost” always on and limited latency for application involving IP or human interaction

k Slide 5 Sub-GHz wireless connectivity platform How to get long range ? there is two way –Go wide band thanks to spread spectrum, and get benefit from diversity and de-spreading gain –Go narrow band and take benefit from –Increased sensivity (less noise) –Limited cost and complexity –Reduced spectrum use (better coexistence) In addition, Frequency Hopping and efficient FEC and interleaving can bring diversity and reduce system margin, even for narrow band system

k Slide 6 Sub-GHz wireless connectivity platform Ultra-low power design to reach multi-year operation : The best way to save power : just do nothing ! But we also need to save bidirectionnality and limited latency Network synchronization : Allow endpoint sto sleep as soon as no activity is planned for them Minimize unwanted wake up and coordinate RX windows : each endpoint stay reachable in acceptable time (always-on illusion) Very short media probing at regular interval Keep network probing duration as short as possible (direct impact on the duty cycle!) and as low level as possible : Full wake up occurs only if some activity is detected on the channel Probing period can be in the order of one to a few second (low latency)

k Slide 7 Sub-GHz ISM license free bands 915MHz, 868MHz, MHz better propagation properties and less interferences than 2.4 Ghz Simple to implement modulation : GFSK / FSK Simple and low power : very power efficient implementation are already available for endpoint Concentrator can offer more complex receiver (coherent, soft decision..) Low data rate and narrow bandwidth : 15 Kbps GFSK modulation 40 KHz bandwidth and 50 kHz channelization Limited noise bandwidth : dBm thermal noise RX sensitivity up to – 115 dBm (endpoint) Low spectrum occupancy : better coexistence properties Fundamentals – RF features

k Slide 8 Intra Frame Frequency Hopping : Provide channel diversity –Frame is sliced into several data blocks –Data block length can be flexible (recommended : 255 symbols) –Hopping occurs between each data block over a N-hopping sequence –Short training symbol at the beginning of each data block –up to 63 channels in EU ( MHz) –x50 channels or less to comply with FCC part (915 MHz) Fundamentals – Reliable Comms SHR + PHR PSDU Frame Chan. #1 Chan. #2 Chan. #N SHR + PHR PSDU – subpart #2 PSDU – subpart #N Data block #1 Data block #2Data block #N PSDU

k Slide 9 Intra Frame Frequency Hopping : Provide channel diversity –First data block include SHR and PHR. –First data block can have increased duration if needed (preamble sampling) –R-Sync : 2 bits “00” for carrier re-sync after hopping –Proposed data block length is : 255 symbols ( ~20 ms) + R-Sync –Only complete data block can be send : add padding bit if required –Minimum frame duration : 2 Data blocks Fundamentals – Reliable Comms SHR + PHR PSDU Frame Chan. #N PSDU – subpart #N R-Sync Data block #1 Data block #2Data block #N

k Slide 10 2 FEC for asymmetric protection : Generic use : Convolutional codes K=7 (171,133) Already included in PHY Endpoint : efficient protection for reasonable complexity (dmin=10) Gateway : more complex RX implementation allow to increase performance by 4 dB (hard decision, coherent RX) Uplink : Endpoint to Gateway -> can use Turbo Code Simple encoding is done by the endpoint – more complex decoding is handled by Gateway. Gateway is able to decode both FEC, whatever endpoint selected Will provide close to Shannon limit performance. Fundamentals – Reliable Comms

k Slide 11 Data interleaving + LSFR data scrambling (whitenning) Spread data and code bit among data blocks Each blocks carry data bits and uncorrelated code bits Fundamentals – Reliable Comms Random Interleaver : Optimal performance require 7 data blocks (typical 100 bytes frame) But some frame will be shorter. Random interleaver would allow Variable size between 2 to 7 Data blocks. Uniform distribution.

k Link Budget – 915 MHz Slide 12 Channel Model ParametersNotes Frequency (MHz)915Valid Range MHz Collector Antenna Height (m)30Valid Range m, including terrain. Endpoint Antenna Height (m)1Valid Range 1-10 m Distance (km)2Valid Range 1-20 km Downlink Path Loss CalculationNotes Collector Tx Power (dBm)30Subject to Tx Power Regulations Collector Tx Antenna Gain (dBi)6Subject to Tx Power Regulations Hata Path Loss (dB)-128,53 Must reference the right path loss from the next worksheet Shadowing Margin (dB)-16To buffer against variable shadowing loss Penetration Loss (dB)-10For underground vaults, etc. Endpoint Rx Antenna Gain (dBi)2 If using same antenna for Tx, must be same as in Uplink Table Endpoint Interference (dB)1Rise over Thermal Interference Rx Power at Endpoint (dBm)-115,53Compare against Rx sensitivity Uplink Path Loss CalculationNotes Endpoint Tx Power (dBm)27 Subject to Tx Power Regulations. Can be different from Collector Endpoint Tx Antenna Gain (dBi)2Subject to Tx Power Regulations Penetration Loss (dB)-10For underground vaults, etc. Hata Path Loss (dB)-128,53Same as Downlink Shadowing Margin (dB)-16Same as Downlink Collector Rx Antenna Gain (dBi) 6If using same antenna for Tx, must be same as in Downlink Table Collector Interference (dB)10Rise over Thermal Interference Rx Power at Collector (dBm)-109,53Compare against Rx sensitivity Scenario 1 : 2 km range in a Sub-urban area, between pole and indoor meter Indoor endpoint

k Link Budget – 868 MHz 25 mW Slide 13 Help to save endpoint power Scenario 2 : 1 km range in a Sub-urban area, between pole and typical outdoor gaz meter Outdoor endpoint Channel Model ParametersNotes Frequency (MHz)868Valid Range MHz Collector Antenna Height (m)30Valid Range m, including terrain. Endpoint Antenna Height (m)1Valid Range 1-10 m Distance (km)1Valid Range 1-20 km Downlink Path Loss CalculationNotes Collector Tx Power (dBm)14Subject to Tx Power Regulations Collector Tx Antenna Gain (dBi)0Subject to Tx Power Regulations Hata Path Loss (dB)-117,47 Must reference the right path loss from the next worksheet Shadowing Margin (dB)-16To buffer against variable shadowing loss Penetration Loss (dB)0For underground vaults, etc. Endpoint Rx Antenna Gain (dBi)2 If using same antenna for Tx, must be same as in Uplink Table Endpoint Interference (dB)1Rise over Thermal Interference Rx Power at Endpoint (dBm)-116,47Compare against Rx sensitivity Uplink Path Loss CalculationNotes Endpoint Tx Power (dBm)12 Subject to Tx Power Regulations. Can be different from Collector Endpoint Tx Antenna Gain (dBi)2Subject to Tx Power Regulations Penetration Loss (dB)0For underground vaults, etc. Hata Path Loss (dB)-117,47Same as Downlink Shadowing Margin (dB)-16Same as Downlink Collector Rx Antenna Gain (dBi)6 If using same antenna for Tx, must be same as in Downlink Table Collector Interference (dB)10Rise over Thermal Interference Rx Power at Collector (dBm)-103,47Compare against Rx sensitivity

k Link Budget – 868 MHz 25 mW Slide 14 Scenario 3 : 100m range in a Sub-urban area, between a rooftop device and underground water meter (typical relaying scenarios) Underground endpoint Channel Model ParametersNotes Frequency (MHz)868Valid Range MHz Collector Antenna Height (m)20Valid Range m, including terrain. Endpoint Antenna Height (m)1Valid Range 1-10 m Distance (km)0,1Valid Range 1-20 km Downlink Path Loss CalculationNotes Collector Tx Power (dBm)12Subject to Tx Power Regulations Collector Tx Antenna Gain (dBi)2Subject to Tx Power Regulations Wallfish Ikegami Path Loss (dB)-99,57 Must reference the right path loss from the next worksheet Shadowing Margin (dB)-10To buffer against variable shadowing loss Penetration Loss (dB)-20For underground vaults, etc. Endpoint Rx Antenna Gain (dBi)2 If using same antenna for Tx, must be same as in Uplink Table Endpoint Interference (dB)1Rise over Thermal Interference Rx Power at Endpoint (dBm)-112,57Compare against Rx sensitivity Uplink Path Loss CalculationNotes Endpoint Tx Power (dBm)12 Subject to Tx Power Regulations. Can be different from Collector Endpoint Tx Antenna Gain (dBi)2Subject to Tx Power Regulations Penetration Loss (dB)-20For underground vaults, etc. Hata Path Loss (dB)-99,57Same as Downlink Shadowing Margin (dB)-10Same as Downlink Collector Rx Antenna Gain (dBi)2 If using same antenna for Tx, must be same as in Downlink Table Collector Interference (dB)5Rise over Thermal Interference Rx Power at Collector (dBm)-108,57Compare against Rx sensitivity Wallfish Ikegami model as valid range down to 20m.

k Frame Fragmentation Transmitting long frame at a low date rate can be problematic : Channel coherence time is estimated to 20 ms Frame error rate increased when frame length increase But Fast FH provide a «de-facto» PHY fragmentation : A frame is sliced in several “data blocks” (min 2 / max 128) Data block duration is shorter than channel coherence time Frame length and data block size give the number of data block Each data block is identified by his position in the hopping sequence Only ACK need to be modified to allows signaling of damaged data block Slide 15

k Slide e TSCH resources management Time Slot Channel Hopping defines the automatic repetition of a slotframe based on a shared notion of time TSCH Allows the devices hopping over the entire channel space in a slotted way thus minimizing the negative effects of multipath fading and interference while avoiding collisions Slotframe is configurable through the definition of the channels used, the number of slots and the duration of the slots TSCH parameters will define data block duration MAC Layer compatibility

k Slide 17 Packet fragmentation in a slotframe Fast FH provide a PHY level fragmentation The adaptation layer between PHY and MAC uses the channel management scheme of 15.4e TSCH mode for hopping on PHY data blocks The first data block carries the PHR, including frame length and thus the total number of data blocks. The following data blocks are sent on the different channels. After the last data block, a group ACK is sent in the other direction, each bit of this group representing the correct reception of the corresponding data block. Only the data blocks which have not been correctly received are retransmitted. MAC Layer compatibility

k Slide 18 As an example, with a 15kbps data rate and ½ FEC, 16 data bytes can be transmitted in a ~20ms slot  Data block duration is shortest than channel coherence time  128 slots are required for transmitting 2047 bytes (longuest possible frame)  Average 100 bytes frames will requires 7 data blocks Example

k Slide 19 Several mean to help coexistence -Narrow channel (50 kHz) : limited spectrum usage -Frequency hopping : -Adequate sequences management mitigate interference between independent networks -Short data block minimize interference on a single channel as individual channel occupancy time remain low Coexistence

k Conclusions -FSK/GFSK are proven solutions : -To address very low power wireless devices -To allow flexible implementations -Narrow band, FH, efficient FEC and interleaving allow supporting path loss larger than 140 dB. -Frequency Hopping bring channel diversity and frame fragmentation “built-in” : improved robustness -Relaying allows yet improved network coverage and network resilience against major channel changes -but handling yet higher path loss requires other technology -Limited latency and always-on behaviour can be provided at an acceptable cost Will be happy to discuss exchange with everybody interested Slide 20

k Slide 21 Thank You

k FSK Receiver implementation Slide 22 Comparison between : - Low cost non coherent FSK receiver using hard decision and viterbi decoder - Coherent FSK receiver using soft decision and viterbi decoder - Performance increase by 4 dB at BER = 1.10 e -3