Internet of Things (IoT): Wide-area IoT protocols*

Transcription

1 Internet of Things (IoT): Wide-area IoT protocols* *Long-range IoT radio access technologies (RATs) Tampere University of Technology, Tampere, Finland Vitaly Petrov:

2 People-centric IoT applications and services

3 Features of long-range IoT radio access technologies (RATs) q Massive amount of connected devices (we currently talk about ~ devices per square meter) q Specific traffic characteristics q Stringent requirements of the end devices Energy consumption / battery lifetime Communication range Simplicity of the solution / low cost q Therefore, we end up designing dedicated RATs for IoT, preferably, long-range.

4 Historical insights.. q Originally, there was no common understanding that dedicated technologies are needed. q Approach 1: Let machines talk conventional Wi-Fi q Approach 2: Let machines talk conventional LTE q Below, we discuss, why this idea is not nice at all

5 Approach 1 motivation (Conventional Wi-Fi for IoT) App App App App Interoperability, connectivity, access control, service discovery, privacy IoT optimized fixed/wireless connectivity q Mitigate technology fragmentation by reusing existing deployments

6 Problem statement q Wi-Fi de-facto technology for WLANs q Expected to minimize time-to-market for various MTC applications q Assessing their performance is important *N. Hunn, Using for M2M, stories/hunnezurio.pdf

7 Network architecture

8 Challenges q Application challenges Small burst transmissions Large number of devices Tight battery budget q Technology challenges Contention-based access High signaling overhead

9 Assessment methodology

10 IEEE test bench Measurement of PER 1. MTC aware use case 2. State-of-the-art technology 3. Saturated queues 4. Two signaling schemes Open-source driver: ath9k Manual rate control No retries (genuine statistics)

11 PER measurements Basic scheme RTS/CTS scheme q Realistic levels of PER estimated

12 Goodput prediction Optimistic scenario: Full-buffer traffic Single link Fixed data rate Maximum power

13 Goodput estimation Basic scheme RTS/CTS scheme Semi-analytical model verified *G. Martorell, F. Riera-Palou, and G. Femenias, Closed-Loop Adaptive IEEE n with PHY/MAC Cross-Layer Constraints, MACOM 2011

14 Goodput accuracy Acceptable accuracy level

15 Reference MTC device A typical MTC device Traffic rate 100 Kbps Message size ~20 bytes *N. Himayat, S. Talwar, K. Johnsson, S. Andreev, O. Galinina, A. Turlikov, Proposed IEEE p Performance Requirements for Network Entry by Large Number of Devices, IEEE Broadband Wireless Access Working Group C80216p , 2010.

16 Deployment requirements *A. Maeder, P. Rost, D. Staehle, The Challenge of M2M Communications for the Cellular Radio Access Network, 11 th Wurzburg Workshop on IP EuroView2011, 2011 q Wi-Fi coverage area ~50 meters q ~200 MTC devices per cluster

17 Conventional system Conventional system works bad q Scales poorly q Too much overhead

18 Aggregation techniques q MAC layer aggregation q PHY layer aggregation *defined by IEEE *defined by IEEE n-2009

19 Aggregation results (asymptotic) q Potential for some improvement

20 Numerical results (optimistic) q Scalable MTC support is challenging

21 Approach 1. Major conclusions q Main results Conventional IEEE scales poorly for MTC IEEE n-2009 aggregation looks promising, but still non-sufficient

22 Approach 2. Use conventional LTE for IoT. System topology Aggregated traffic (M2M applications) Traffic analysis (numerous devices) Performance metrics analysis (closed-form approximation) Throughput Mean delay Energy efficiency Target: Performance evaluation Challenges Overload protection Energy efficiency Small data transmission 22

23 Approach 2. Initial network entry UE enodeb UE enodeb Traffic arrival Delay SG Rx + processing Scheduling opportunity Scheduling request Scheduling grant (SG) Data transmission Scheduling opportunity Periodicity Delay Traffic arrival Response window + backoff window RAR Rx + processing Ramping failure Ramping failure Msg1: Preamble Msg1: Preamble retransmission Msg2: Random access response (RAR) Msg3: Layer 3 message Msg4: Contention resolution identity Collision

24 LTE RACH model Specialty: dependant queues ~30000 sources 54 preambles Metrics (closed-form approximation) throughput energy expenditure energy efficiency 24

25 Approach 2 numerical results (1) q Delay is too high for all the schemes Probability COBALT PRACH PUCCH Delay, ms

26 Approach 2 numerical results (2) q Energy efficiency is too low for all the schemes... Individual power consumption, mw PUCCH PRACH COBALT Theory Number of MTC devices

27 Approach 2. Major conclusions q Main results Conventional LTE scales poorly for MTC LTE-Advanced with certain channel access enhancements scales better, but still not sufficient

28 Therefore, novel IoT-specific radio access technologies are proposed q Due to the abovementioned limitations of conventional RATs for IoT, novel IoT-specific radio access technologies were proposed q These are, including but not limited to: SIGFOX LoRaWAN IEEE ah (branded as Wi-Fi HaLow) NB-IoT (roughly, IoT version of LTE)

29 SIGFOX

30 LoRaWAN

31 Wi-Fi HaLow (IEEE ah)

32 Narrow band IoT (NB-IoT)

33 Motivation to go lower in carrier and bandwidth q Why do we move from 2.4GHz to 900MHz? q Why do we move from 5MHz channel bandwidth in LTE to 180kHz channel bandwidth in NB-IoT?

34 Motivation to go lower in carrier and bandwidth q Why do we move from 2.4GHz to 900MHz? Answer: lower path loss -> greater comm. range q Why do we move from 5MHz channel bandwidth in LTE to 180kHz channel bandwidth in NB-IoT?

35 Motivation to go lower in carrier and bandwidth q Why do we move from ~2GHz to ~900MHz? Answer: lower path loss -> greater comm. range q Why do we move from 5MHz channel bandwidth in LTE to 180kHz channel bandwidth in NB-IoT? Answer: higher Tx power spectral density -> greater comm. range

36 Long-range IoT technologies comparison

37 Long-range IoT RATs vs short-range IoT RATs q Long-range strengths: More devices connected to a single access point -> economical gains Greater communication range -> more applications/services supported... q Short-range strengths: Ability to control/modify internal parameters and the topology -> Greater level of flexibility Shorter communication range -> higher security...