IoT in LTE

Agenda1. Introduction to IOT

Iot NB-Iot LTE -M

2. Requirements, Challenges and solutions Coverage Power Saving Cost Congestion MCS Selection

What is IoT?

IoT is a network of devices embedded with sensors and network connectivity that enables them to collect and exchange data.

How it works?

Collection of data Transmission of selected data through a

communication network Assessment of the data Response to the available information

Example: Smart Parking

Collection of data -System relies on sensors embedded in the parking area.

Transmission of selected data through a communication network – When a car parks on the sensor, it is detected and the sensor

relays the information wirelessly to the gateway. Gateway sends the information through network to the database.

Assessment of the data – Information is updated in the database and a central control can

perform analytics about the parking area occupancies according to time and day.

Response to the available information – Occupancy is reported to users via apps and panels in the street.

M2M Architecture

IoT Standards Development

3GPP has been working on 3 different IoT standard solutions-•LTE-M based on LTE evolutions, Cat0(rel 12) and cat-1(rel 13)•EC-GSM – A narrowband solution based on GSM evolution, •NB-LTE- A narrowband cellular IoT solution, also known as clean state solutions, Cat200KHz.

Later, EC-GSM and NB-LTE, were combined for standardization as a single NB-IoT technology.NB-IoT would support three modes of operation :-

1. ‘Stand-alone operation’ utilizing, for example, the spectrum currently being used by GERAN systems. i.e. replacing a GSM carrier with NB-IoT carrier.2. ‘Guard band operation’ utilizing the unused resource blocks within a LTE carrier’s guard-band, and3. ‘In-band operation’ utilizing resource blocks within a normal LTE carrier.

A Comparison

Requirements & Challenges

Cheap devices => Provisioning of low cost & low complexity UE’s.

Battery Life > 10 yrs => UE Power Consumption Optimization.

MTC devices installed in basements or covered by insulation leading to penetration losses => Coverage Improvement of 20dB is targeted as compared to legacy LTE(category 1 UE- 142.7 dB).

Millions of connected devices expected => Congestion Control.

Low data rate requirements => choosing appropriate MCS.

Solutions

Battery Life solution:UE Power Consumption Optimization(UEPCOP))

4 solutions have been proposed by 3GPP to lower UE’s power consumption-1. Extended DRX cycle in IDLE mode.2. Extended DRX cycle in CONNECTED

mode.3. Power Saving Mode.4. Reducing the RF Bandwidth

Basics of Discontinous Reception(DRX)

Mechanism to save UE energy UE keeps it’s receiver circuitry

off for certain time period i.e. it goes into sleep and wake up states.

In wake up state it listens to PDCCH whereas in sleep state it’s receiver is off.

Terminology

ON Period – Period during which UE should monitor PDCCH. OFF Period – UE enters sleep state. DRX Cycle – 1ON + 1 OFF Period. DRX Inactivity Timer- After reception of PDCCH in ON period, UE

starts this timer and monitors PDCCH in every consecutive subframe till it expires.

DRX Retransmission Timer- When data on PDSCH for a HARQ process is nacked by UE, it knows that it will be retransmitted by eNB, so it starts this timer during which it’ll not go into sleep mode and listen to PDCCH continously for the retransmission.

DRXStartOffset  - Subframe when DRX cycle starts.

DRX short cycle is optional and is for applications like VOIP that require transmission of small amount of data at short, regular intervals.

Types

DRX can be implemented in IDLE as well as Connected mode. In Idle mode, DRX is same as paging cycle In Connected mode, UE needs to listen to PDCCH for

data scheduling.

Basic Procedure

UE is in RRC Connected mode and is continuously monitoring PDCCH. At this point, there is DL Grant and downlink data. The DRX inactivity timer and the main RRC Inactivity timer are restarted

There is UL grant for UE. With DL Grant both DRX and RRC inactivity timers are restarted. 4 ms later UE sends data in uplink

The DRX Inactivity timer is expired since there were no further grants in uplink or downlink. Though UE was constantly monitoring PDCCH. UE now enters the short DRX cycle. The battery savings have just started

 The DRX short cycle timer got expired therefore UE will end up its short DRX cycle and enter the long DRX cycle

The main RRC inactivity timer got expired since there was no activity in uplink or downlink for the duration for RRC Inactivity timer. The UE will go to RRC IDLE state. In idle state UE will use paging DRX cycle

How is it configured?

DRX in IDLE mode can be cell specific or UE specific. Default DRX cycle or cell specific DRX cycle is

configured at eNB and broadcasted in SIB2 to ue’s Dedicated or UE specific DRX value is indicated by

UE to MME in attach request . Both eNB and UE uses smaller of the 2 values. Only On-duration and inactivity timer are signalled

to UE by eNB.

Extended DRX cycle in IDLE mode On increasing DRX cycle length in idle mode i.e.

paging cycle length, UE can go to sleep for longer duration but can remain attached.

Issues and impact

Issue1- Now, both eNB and UE adopt UE specific extended

DRX value. This can impact SIB reading-Suppose extended DRX cycle length is longer than

modification period. In this case, eNB would change SI and start broadcasting SIB’s but UE would not listen. In legacy LTE, BCCH modification period is a multiple of default DRX value. Hence, UE could read modified sib’s.

Solution-Perform cell search and SI reading before the active time.

Power calculations

R2-132394 proposes a power consumption model with following parameters-

P one normal DRX cycle = Tsleep * Psleep + (Tprepare + Ton-duty) * PrxTsleep = DRX cycle – Tprepare – Ton-dutye.g. when DRX cycle is 2.56sec,P one normal DRX cycle = (2560 – 34 – 1) * 0.01 + (34 + 1) * 1 = 60.25In case of extended DRX (> BCCH modification period):Pone extended drx cycle = Tsleep * Psleep + (Tprepare + Ton-duty + TSI-reading + Tcell-detection) * PrxTsleep = Extended DRX cycle – Tprepare – Ton-duty – TSI-reading – Tcell-detectionTprepare = 0 ; coz it would be taken care of in Tcell-detection.e.g. when extended DRX cycle is 10.24sec (= one System Frame),P one extended drx cycle = (10240–1–200–600) * 0.01 + (1+200+600) * 1 = 895.39

For the power consumption comparison between the normal DRX case and the extended DRX case, above P one normal drx cycle should be multiplied by (Extended DRX cycle ÷Normal DRX cycle).Ex, when the extended DRX is 10.24sec and the normal DRX is 2.56sec, the power consumption for the normal DRX shall be multiplied by 4.

Gain by the Extended DRX can be seen when the Extended DRX is longer than 6 system frame length and from there the more the DRX is extended the more gain can be observed.

Extended DRX cycle in CONNECTED mode

It refers to increasing the DRX cycle length in connected mode, similar to idle mode extension.

Issues and impacts

The device should be delay tolerant since, extending the cycle length would mean delay in dl data.

If there is no activity in ul and dl, RRC inactivity timer expires and UE goes into idle mode and RRC connection is released. We need to increment it’s value.

When long DRX cycle is used, re transmissions on higher layers needs to be minimized. Without adjusting the retransmission timers, longer DRX upto several minutes may impact the reachability of the UE. For ex, when TCP sender sends data to an MTC device, it starts a retransmission timer and if it doesn't receive ack till it's expiry it'll retransmit and in worst case it would be unable to transmit it.

Power Saving Mode(PSM)

After a UE goes into idle mode, it releases RRC connection. It then starts an active timer  while performing all idle mode functions i.e. PLMN selection, cell selection/reselection, paging. When active timer expires, UE enters into PSM and starts a Periodic Update Timer, expiry of which will indicate the end of PSM. In this time, the UE stops reading paging or performing  any AS(cell/PLMN selection) or NAS(MM procedures) functions. Also, network should not send any data or page the device.

When UE wants to use PSM, it'll request an active timer in attach/TAU request. If eNB supports PSM it'll select a timer value from the UE given value or MME given configuration. Afterwards, if UE wants to change it's value due to certain condition changes, it'll request the value in TAU procedure

Max duration of PSM is 12.1 days.

The following image is retrieved from http://www.sharetechnote.com/.

Issues and Impacts •A transition within IDLE state is required.

•Most networks today expect contact with device every 2-4 hours, otherwise it  is considered as not reachable and it quietly detaches it from the network so that it limits the devices it needs to keep track of. But now, n/w needs to maintain a very large number of devices that will contact the n/w in a week or two.An optimization at eNB would be to decide a PO within the active time period. When eNB receives a paging message from MME, it finds a PO within active time period, if it is unable to find such PO it'll stop paging at RAN.

E-DRX vs PSM

Extended DRX cycle Power Saving State for devices

Applicability - UEs that can always tolerate traffic with longer access delays for MT services,

- Relatively infrequent data.

Same as for ‘extended DRX cycle’

Power saving gain s Reduction of :- paging monitoring- measurements

Reduction of :- paging monitoring- measurements- SIB monitoring- cell/RAT/PLMN selection - MM procedures

Impact on mobility - Supports mobility- Cell reselection may suffer

latency due to possibly reduced frequency of measurements

- No mobility support when UE is in power saving state (UE executes cell selection when leaves the power saving state)

Impacts to specification - Potential modification to paging if DRX Cycle is extended beyond 1024 radio frames in LTE or 4096 radio frames in HSPA

- Updates to RRM requirements may be necessary (RAN4 performance requirements may need to be updated for longer DRX cycles).

- UE and eNB capability support

- No new RRC state is needed, transitions within the RRC idle state need to be defined (including a new timer and corresponding NAS signalling).

- Negotiation/confirmation of UE/network capability of this functionality

Coverage Enhancement Solutions

Some MTC UE’s are installed in basement or covered by insulation, leading to penetration losses

Coverage improvement of 20dB is required compared to category 1 UE.

Basic Terminology – What is Coverage?

Coupling Loss –Total channel loss over the link b/w UE antenna ports and eNB antenna

ports, Includes path loss. MCL-

Limit value of coupling loss at which service can be delivered and defines the coverage of the service.

Receiver Sensitivity-It is a range  of power within which receiver can listen.

Sender's TX power - Loss = Receiver's Sensitivity.=> UL MCL = UL max TX Power - eNB SensitivityAnd DL MCL = DL max TX Power - UE Sensitivity

TS 36.824 proposes an MCL calculation template as follows-

Physical channel name Value

Transmitter

(1) Tx power (dBm)

Receiver

(2) Thermal noise density (dBm/Hz)

(3) Receiver noise figure (dB)

(4) Interference margin (dB)

(5) Occupied channel bandwidth (Hz)

(6) Effective noise power = (2) + (3) + (4) + 10 log(5) (dBm)

(7) Required SINR (dB)

(8) Receiver sensitivity = (6) + (7) (dBm)(9) MCL = (1) (8) (dB)

3 Solutions have been proposed by 3gPP to enhance coverage

1. TTI Bundling Enhancement2. Repetitions3. PSD Power Boosting

Repetitions:

Repetitions means to transmit a TB in more than 1 sf's.

2 Coverage Enhancement (CE) modes are defined- Mode 1 describes behaviors agreed for no repetitions

or small number of repetitions. Mode2 describes behaviors agreed for large number

of repetitions. A CE level is configured for all channels in a UE.  One CE level can be configured with a set of

repetition numbers at least for PDSCH, PUSCH & MPDCCH

Impact & Issues

Allowing repetition for channels can have an impact on how the channels are allocated.

For ex, PUCCH resource determination from ECCE linkage assumes fixed timing gap b/w PUCCH and PDSCH, but due to different number of repetitions, it could vary, leading to resource collision.

Basics of TTI Bundling

If we talk about a normal uplink TB, it is converted into various redundancy versions and first redundancy version is sent in a subframe followed by consecutive transmissions based on HARQ result.

TTI Bundling is a mechanism in which UE transmit a PUSCH (with different redundancy versions) all in multiple subframes continously. i.e. UE sends the same packet but with different error correction and detection bits in consecutive sf's w/o waiting for HARQ result. The retransmission of TTI bundle is also a TTI bundle.

Why TTI Bundling

To increase the decoding possibilty of TB for cell edge UE's where the coverage is poor. Coz if we know coverage is poor, then instead of waiting for HARQ nack's, we can transmit all RV's together, decoding which would be better.

On increasing redundancy, SINR would decrease leading to greater MCL.

It saves a lot of signalling overhead as only 1 PDCCH assignment is needed for multiple transmissions. Also, it sends ACK/NACK for the entire bundle rather than every retransmission.

1 RLC header is used for all versions and 1 HARQ process ID.

Figures-

1. Current supported TTI Bundle size- 42. Max allowed PRB per sf - 3

Scope of further enhancement for MTC UE's-

1. Introduction of a new TTI Bundle size,2. Introduction of a new RTT and HARQ timing for a bundle size3. Allow more than 3 PRB allocated per subframe.

Introduction of a new TTI Bundle size-

If we increase the bundle size, decoding possibility increases but at the expanse of more PRB resource usage and slightly longer delay.

It has been experimented that on increasing bundle size from 4 to 8 approx 3DB gain can be expected.

What should be max Bundle size? Inter arrival time of VOIP packets is 20msec i.e.

AMR codec provides packets at 20 msec interval. Hence, it was a natural choice to limit the max number of TTI for packet to be 20. However, since majority of packets use less than max TTI with 2% bler target, it will not result in queue build up.

For ex, 8 TTI bundling with max 3 HARQ transmissions(upto 24 TTI's per VOIP packet) provided a 1DB gain over 4 TTI Bundle with 4 HARQ retransmissions.

3GPP has not concluded a figure for this yet.

Introduction of a new RTT and HARQ timing for a bundle size-

Reduced round trip time can bring benefits for more energy accumulation for VoIP within given delay budget. However, it will become difficult for scheduler to effectively cohandle rel 8/9/10 UE's

Allow more than 3 PRB allocated per subframe- This allows flexibility in UE scheduling. Limiting it

to 3 PRB's  may not allow UE to take full advantage of dynamic change in channel and limit it's performance.

Issues

It is possible that the packet is decoded in the first transmission, then it leads to wastage of time and more energy consumption than needed.

Provisioning of Low Cost & ComplexityWe’ll discuss 3 solutions for reducing the cost and

complexity of a device-1. Reduction Of Maximum Bandwidth.2. Reduction of supported downlink transmission

modes.3. Reduction of transmit power.

Reduction Of Maximum Bandwidth.

It has been experimented by various companies and concluded that upto 40-50% cost savings can be achieved by reducing bandwidth.

Reducing the max bandwidth from 20MHz to 1.4 MHz has been proposed.

There are 3 cases- Reduce bandwidth for both RF and Baseband Reduce bandwidth for baseband only Reduce bandwidth for baseband for data channel

only

Impacts and Issues

For all the options, coverage of PDSCH and PUSCH can be affected because of loss in frequency selective scheduling.

Power consumption is minimized as discussed earlier. However, due to degradation in PDSCH and PUSCH, transmission time may become longer leading to increase in power consumption.

Reduction of Transmit Power Removing the power amplifier stage of an MTC UE

is expected to provide cost benefits. Impact&Issues-

UL coverage will degrade

Reduction of supported downlink TM Modes For rel10 , cat-1 UE’s there are 9 TM modes TM1-

TM9. For IoT, it is limited to TM1 and TM2.

Reduced data rates

For cat0 and cat-m UE’s Dl/ul- 1Mbps

For NB-IoT UE’s Dl- 200Kbps Ul-144Kbps

Data rate calculation

LTE-MData rate =( no. of RB ‘s * no. of subcarriers in an RB

* no. of symbols in a subcarrier * no. of slots * modulation order ) bits/msec

= 6*12*7*2*4 = 4002 bits/msec = 4 Mbps25% of it would be used for control signalling=> data rate = 1 Mbps.

MCS and TBS

64QAM is not supported for MTC LTE UE’s. Due to increase in redundacy; code rate, SINR and

efficiency would decrease. CQI entries with 64QAM are replaced with entries

representing lower SINR. Minimum TB size is 16bits and max is 680 bits. Hence, 4 bit MCS table would be sufficient. Assuming CE level would be changed into a lower one,

UE may report CQI index corresponding to out-of-range, which may result in no DL PDSCH transmission because of no matched CQI value with current channel condition. In this case, we can define CQI indices representing lower spectral efficiency values, which avoid the situation what we discussed above.

CQI index Modulation code rate x 1024 x efficiency

0 out of range

1 QPSK 40 0.0781

2 QPSK 78 0.1523

3 QPSK 120 0.2344

4 QPSK 193 0.3770

5 QPSK 308 0.6016

6 QPSK 449 0.8770

7 QPSK 602 1.1758

8 16QAM 378 1.4766

9 16QAM 490 1.9141

10 16QAM 616 2.4063

11 Reserved Reserved Reserved

12 Reserved Reserved Reserved

13 Reserved Reserved Reserved

14 Reserved Reserved Reserved

15 Reserved Reserved Reserved

MCS table for PUSCH

MCS Index Modulation Order TBS Index

0 2 0

1 2 1

2 2 2

3 2 3

4 2 4

5 2 5

6 2 6

7 2 7

8 2 8

9 2 9

10 2 10

11 4 10

12 4 11

13 4 12

14 4 13

15 4 14

References TR 36.888 TR 37.868 TR 36.824 R2-132609 R2-132560 R2-131793 R2-132395 R2-132394 R2-132380 R2-132893 TR 43.869 TS 36.211 TR 37.868 R1-154719 TR 23.770