idnits 2.17.1 draft-johansson-rmcat-scream-cc-04.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == The document doesn't use any RFC 2119 keywords, yet has text resembling RFC 2119 boilerplate text. -- The document date (January 12, 2015) is 3393 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-02) exists of draft-sarker-rmcat-cellular-eval-test-cases-00 == Outdated reference: A later version (-13) exists of draft-ietf-tcpm-newcwv-07 Summary: 0 errors (**), 0 flaws (~~), 4 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 RMCAT WG I. Johansson 3 Internet-Draft Z. Sarker 4 Intended status: Informational Ericsson AB 5 Expires: July 16, 2015 January 12, 2015 7 Self-Clocked Rate Adaptation for Multimedia 8 draft-johansson-rmcat-scream-cc-04 10 Abstract 12 This memo describes a rate adaptation algorithm for conversational 13 video services. The solution conforms to the packet conservation 14 principle and uses a hybrid loss and delay based congestion control 15 algorithm. The algorithm is evaluated over both simulated Internet 16 bottleneck scenarios as well as in a LTE (Long Term Evolution) system 17 simulator and is shown to achieve both low latency and high video 18 throughput in these scenarios. 20 Status of This Memo 22 This Internet-Draft is submitted in full conformance with the 23 provisions of BCP 78 and BCP 79. 25 Internet-Drafts are working documents of the Internet Engineering 26 Task Force (IETF). Note that other groups may also distribute 27 working documents as Internet-Drafts. The list of current Internet- 28 Drafts is at http://datatracker.ietf.org/drafts/current/. 30 Internet-Drafts are draft documents valid for a maximum of six months 31 and may be updated, replaced, or obsoleted by other documents at any 32 time. It is inappropriate to use Internet-Drafts as reference 33 material or to cite them other than as "work in progress." 35 This Internet-Draft will expire on July 16, 2015. 37 Copyright Notice 39 Copyright (c) 2015 IETF Trust and the persons identified as the 40 document authors. All rights reserved. 42 This document is subject to BCP 78 and the IETF Trust's Legal 43 Provisions Relating to IETF Documents 44 (http://trustee.ietf.org/license-info) in effect on the date of 45 publication of this document. Please review these documents 46 carefully, as they describe your rights and restrictions with respect 47 to this document. Code Components extracted from this document must 48 include Simplified BSD License text as described in Section 4.e of 49 the Trust Legal Provisions and are provided without warranty as 50 described in the Simplified BSD License. 52 Table of Contents 54 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 55 1.1. Wireless (LTE) access properties . . . . . . . . . . . . 3 56 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 3 57 3. Overview of SCReAM Algorithm . . . . . . . . . . . . . . . . 3 58 3.1. Congestion Control . . . . . . . . . . . . . . . . . . . 4 59 3.2. Transmission Scheduling . . . . . . . . . . . . . . . . . 5 60 3.3. Media Rate Control . . . . . . . . . . . . . . . . . . . 5 61 4. Detailed Description of SCReAM . . . . . . . . . . . . . . . 5 62 4.1. SCReAM Sender . . . . . . . . . . . . . . . . . . . . . . 5 63 4.1.1. Constants and Parameter values . . . . . . . . . . . 7 64 4.1.2. Network congestion control . . . . . . . . . . . . . 11 65 4.1.2.1. Congestion window update . . . . . . . . . . . . 12 66 4.1.2.2. Transmission scheduling . . . . . . . . . . . . . 15 67 4.1.3. Video rate control . . . . . . . . . . . . . . . . . 16 68 4.2. SCReAM Receiver . . . . . . . . . . . . . . . . . . . . . 19 69 5. Feedback Message . . . . . . . . . . . . . . . . . . . . . . 20 70 6. Additional features . . . . . . . . . . . . . . . . . . . . . 21 71 6.1. Packet pacing . . . . . . . . . . . . . . . . . . . . . . 21 72 6.2. Frame skipping . . . . . . . . . . . . . . . . . . . . . 21 73 6.3. Q-bit semantics (source quench) . . . . . . . . . . . . . 23 74 7. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 24 75 8. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . 24 76 9. Open issues . . . . . . . . . . . . . . . . . . . . . . . . . 24 77 10. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 25 78 11. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 25 79 12. Security Considerations . . . . . . . . . . . . . . . . . . . 25 80 13. Change history . . . . . . . . . . . . . . . . . . . . . . . 25 81 14. References . . . . . . . . . . . . . . . . . . . . . . . . . 26 82 14.1. Normative References . . . . . . . . . . . . . . . . . . 26 83 14.2. Informative References . . . . . . . . . . . . . . . . . 26 84 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 27 86 1. Introduction 88 Congestion in the Internet is a reality and application that are 89 deployed in the Internet must have congestion control schemes in 90 place not only for the robustness of the service that it provides but 91 also to ensure the function of the currently deployed Internet. As 92 the interactive realtime communication imposes a great deal of 93 requirements on the transport, a robust, efficient rate adaptation 94 for all access type is considered as an important part interactive 95 realtime communications as the transmission channel bandwidth may 96 vary over time. Wireless access such as LTE, which is an integral 97 part of the current Internet, increases the importance of rate 98 adaptation as the channel bandwidth of a default LTE bearer 99 [QoS-3GPP] can change considerably in a very short time frame. Thus 100 a rate adaptation solution for interactive realtime media, such as 101 WebRTC, must be both quick and be able to operate over a large span 102 in available channel bandwidth. This memo describes a solution,named 103 SCReAM, that is based on the self-clocking principle of TCP and uses 104 similar techniques used in a new delay based rate adaptation 105 algorithm, LEDBAT [RFC6817]. Because neither TCP nor LEDBAT was 106 designed for interactive realtime media, a few extra features are 107 needed to make the concept work well within this context. This memo 108 describes these extra features. 110 1.1. Wireless (LTE) access properties 112 [I-D.draft-sarker-rmcat-cellular-eval-test-cases] introduces the 113 complications that can be observed in wireless environments. 114 Wireless access such as LTE can typically not guarantee a given 115 bandwidth, this is true especially for default bearers. The network 116 throughput may vary considerably for instance in cases where the 117 wireless terminal is moving around. 119 Unlike wireline bottlenecks with large statistical multiplexing it is 120 not possible to try to maintain a given bitrate when congestion is 121 detected with the hope that other flows will yield, this because 122 there are generally few other flows competing for the same 123 bottleneck. Each user gets its own variable throughput bottleneck, 124 where the throughput depends on factors like channel quality, network 125 load and historical throughput. The bottom line is, if the 126 throughput drops, the sender has no other option than to reduce the 127 bitrate. In addition, the grace time, i.e. allowed reaction time 128 from the time that the congestion is detected until a reaction in 129 terms of a rate reduction is effected, is generally very short, in 130 the order of one RTT (Round Trip Time). 132 2. Terminology 134 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 135 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 136 document are to be interpreted as described in RFC2119 [RFC2119] 138 3. Overview of SCReAM Algorithm 140 The core SCReAM algorithm has similarities to concepts like self- 141 clocking used in TFWC [TFWC] and follows packet conservation 142 principles. The packet conservation principle is described as an 143 important key-factor behind the protection of networks from 144 congestion [FACK]. 146 The packet conservation principle is realized by including a vector 147 of the sequence numbers of received packets in the feedback, see 148 Section 5, from the receiver back to the sender, the sender keeps a 149 list of transmitted packets and their respective sizes. This 150 information is then used to determine how many bytes can be 151 transmitted. A congestion window puts an upper limit on how many 152 bytes can be in flight, i.e. transmitted but not yet acknowledged. 153 The congestion window is determined in a way similar to LEDBAT 154 [RFC6817]. This ensures that the e2e latency is kept low. The basic 155 functionality is quite simple, there are however a few steps to take 156 to make the concept work with conversational media. These will be 157 briefly described in sections Section 3.1 to Section 3.3. 159 The rate adaptation solution constitutes three parts- congestion 160 control, transmission scheduling and media rate adaptation. All 161 these three parts reside at the sender side. The receiver side 162 algorithm is very simple in comparison as it only generates 163 acknowledgements to received RTP packets. 165 3.1. Congestion Control 167 The congestion control sets an upper limit on how much data can be in 168 the network (bytes in flight); this limit is called CWND (congestion 169 window) and is used in the transmission scheduling. 171 The SCReAM congestion control method, uses LEDBAT [RFC6817] to 172 measure the OWD (one way delay). The SCReAM sender calculates the 173 congestion window based on the feedback from SCReAM receiver. The 174 congestion window is allowed to increase if the OWD is below a 175 predefined target, otherwise the congestion window decreases. The 176 delay target is typically set to 50-100ms. This ensures that the OWD 177 is kept low on the average. The reaction to loss events is similar 178 to that of loss based TCP, i.e. an instant reduction of CWND. 180 LEDBAT is designed with file transfers as main use case which means 181 that the algorithm must be modified somewhat to work with rate- 182 limited sources such as video. The modifications are 184 o Congestion window validation techniques. These are similar in 185 action as the method described in [I-D.ietf-tcpm-newcwv]. 187 o Fast start for bitrate increase. It makes the video bitrate ramp- 188 up within 5 to 10 seconds. The behavior is similar to TCP 189 slowstart. The fast start is exited when congestion is detected. 190 The fast start state can be resumed if the congestion level is 191 low, this to enable a reasonably quick rate increase in case link 192 throughput increases. 194 o Adaptive delay target. This helps the congestion control to 195 compete with FTP traffic to some degree. 197 3.2. Transmission Scheduling 199 Transmission scheduling limits the output of data, given by the 200 relation between the number of bytes in flight and the congestion 201 window similar to TCP. Packet pacing is used to mitigate issues with 202 coalescing that may cause increased jitter and/or packet loss in the 203 media traffic. 205 3.3. Media Rate Control 207 The media rate control serves to adjust the media bitrate to ramp up 208 quickly enough to get a fair share of the system resources when link 209 throughput increases. 211 The reaction to reduced throughput must be prompt in order to avoid 212 getting too much data queued up in the RTP packet queues. The media 213 bitrate is decreased if the RTP queue size exceeds a threshold. 215 In cases where the sender frame queues increase rapidly such as the 216 case of a RAT (Radio Access Type) handover it may be necessary to 217 implement additional actions, such as discarding of encoded video 218 frames or frame skipping in order to ensure that the RTP queues are 219 drained quickly. Frame skipping means that the frame rate is 220 temporarily reduced. Discarding of old video frames is a more 221 efficient way to reduce media latency than frame skipping but it 222 comes with a requirement to repair codec state, frame skipping is 223 thus to prefer as a first remedy. Frame skipping is described as an 224 optional to implement feature in this specification. 226 4. Detailed Description of SCReAM 228 4.1. SCReAM Sender 230 This section describes the sender side algorithm in more detail. It 231 is split between the network congestion control and the video rate 232 adaptation. 234 Figure 1 shows the functional overview of a SCReAM sender. The RTP 235 application interaction with congestion control is described in 236 [I-D.ietf-rmcat-app-interaction]. Here we use a more decomposed 237 version of the implementation model in the sense that the RTP packets 238 may be queued up in the sender, the transmission of these RTP packets 239 is controlled by a transmission scheduler. A SCReAM sender 240 implements rate control and a queue for each media type or source, 241 where RTP packets containing encoded media frames are temporarily 242 stored for transmission, the figure shows the details for when two 243 video sources (a.k.a streams) are used. 245 ---------------------------- ----------------------------- 246 | Video encoder | | Video encoder | 247 ---------------------------- ----------------------------- 248 ^ | ^ ^ | ^ 249 (1)| (2)| (3)| (1)| (2)| (3)| 250 | RTP | | RTP | 251 | V | | V | 252 | ------------- | | ------------- | 253 ----------- | |-- ----------- | |-- 254 | Rate | (4) | Queue | | Rate | (4) | Queue | 255 | control |<----| | | control |<----| | 256 | | |RTP packets| | | |RTP packets| 257 ----------- | | ----------- | | 258 ------------- ------------- 259 | | 260 --------------- -------------- 261 (5)| |(5) 262 RTP RTP 263 | | 264 v v 265 -------------- ---------------- 266 | Network | (8) | Transmission | 267 | congestion |<-------->| scheduler | 268 | control | | | 269 -------------- ---------------- 270 ^ | 271 | (7) |(6) 272 ---------RTCP---------- RTP 273 | | 274 | v 275 ------------- 276 | UDP | 277 | socket | 278 ------------- 280 Figure 1: SCReAM sender functional view 282 Video frames are encoded and forwarded to the queue (2). The media 283 rate adaptation adapts to the size of the RTP queue and controls the 284 video bitrate (1). The RTP packets are picked from each queue based 285 on some defined priority order or simply in a round robin fashion 286 (5). A transmission scheduler takes care of the transmission of RTP 287 packets, to be written to the UDP socket (6). In the general case 288 all media must go through the transmission scheduler and is allowed 289 to be transmitted if the number of bytes in flight is less than the 290 congestion window. Audio frames can however be allowed to be 291 transmitted immediately as audio is typically low bitrate and thus 292 contributes little to congestion, this is something that is left as 293 an implementation choice. RTCP packets are received (7) and the 294 information about bytes in flight and congestion window is exchanged 295 between the network congestion control and the transmission scheduler 296 (8). 298 4.1.1. Constants and Parameter values 300 A set of constants are defined in Table 1, state variables are 301 defined in Table 2. And finally, local variables are described in 302 Table 3. 304 An init value [] indicates an empty array. 306 +-------------------------------+------------------------+----------+ 307 | Constant | Explanation | Value | 308 +-------------------------------+------------------------+----------+ 309 | OWD_TARGET_LO | Min OWD target | 0.1s | 310 | OWD_TARGET_HI | Max OWD target | 0.4s | 311 | MAX_BYTES_IN_FLIGHT_HEAD_ROOM | Headroom for | 1.1 | 312 | | limitation of CWND | | 313 | GAIN | Gain factor for | 1.0 | 314 | | congestion window | | 315 | | adjustment | | 316 | BETA | CWND scale factor due | 0.6 | 317 | | to loss event | | 318 | BETA_R | Target rate scale | 0.8 | 319 | | factor due to loss | | 320 | | event | | 321 | BYTES_IN_FLIGHT_SLACK | Additional slack [%] | 10% | 322 | | to the congestion | | 323 | | window | | 324 | RATE_ADJUST_INTERVAL | Interval between video | 0.1s | 325 | | bitrate adjustments | | 326 | FRAME_PERIOD | Video coder frame | | 327 | | period [s] | | 328 | TARGET_BITRATE_MIN | Min target_bitrate | | 329 | | [bps] | | 330 | TARGET_BITRATE_MAX | Max target_bitrate | | 331 | | [bps] | | 332 | RAMP_UP_TIME | Timespan [s] from | 10s | 333 | | lowest to highest | | 334 | | bitrate | | 335 | PRE_CONGESTION_GUARD | Guard factor against | 0.0..0.2 | 336 | | early congestion | | 337 | | onset. A higher value | | 338 | | gives less jitter | | 339 | | possibly at the | | 340 | | expense of a lower | | 341 | | video bitrate. | | 342 | TX_QUEUE_SIZE_FACTOR | Guard factor against | 0.0..2.0 | 343 | | RTP queue buildup | | 344 +-------------------------------+------------------------+----------+ 346 Table 1: Constants 348 +-------------------------+--------------------+--------------------+ 349 | Variable | Explanation | Init value | 350 +-------------------------+--------------------+--------------------+ 351 | owd_target | OWD target | OWD_TARGET_LO | 352 | owd_fraction_avg | EWMA filtered | 0.0 | 353 | | owd_fraction | | 354 | owd_fraction_hist | Vector of the last | [] | 355 | | 20 owd_fraction | | 356 | owd_trend | OWD trend, | 0.0 | 357 | | indicates | | 358 | | incipient | | 359 | | congestion | | 360 | owd_norm_hist | Vector of the last | [] | 361 | | 100 owd_norm | | 362 | mss | Maximum segment | 1000 | 363 | | size = Max RTP | | 364 | | packet size [byte] | | 365 | min_cwnd | Minimum congestion | 2*MSS | 366 | | window [byte] | | 367 | in_fast_start | True if in fast | true | 368 | | start state | | 369 | cwnd | Congestion window | min_cwnd | 370 | | [byte] | | 371 | cwnd_i | Congestion window | 1 | 372 | | inflection point | | 373 | bytes_newly_acked | The number of | 0 | 374 | | bytes that was | | 375 | | acknowledged with | | 376 | | the last received | | 377 | | acknowledgement | | 378 | | i.e. bytes | | 379 | | acknowledged since | | 380 | | the last CWND | | 381 | | update [byte]. | | 382 | | Reset after a CWND | | 383 | | update | | 384 | send_wnd | Upper limit of how | 0 | 385 | | many bytes that | | 386 | | can be transmitted | | 387 | | [byte]. Updated | | 388 | | when CWND is | | 389 | | updated and when | | 390 | | RTP packet is | | 391 | | transmitted | | 392 | t_pace | Approximate | 0.001 | 393 | | estimate of inter- | | 394 | | packet | | 395 | | transmission | | 396 | | interval [s], | | 397 | | updated when RTP | | 398 | | packet transmitted | | 399 | age_vec | A vector of the | [] | 400 | | last 20 RTP packet | | 401 | | queue delay | | 402 | | samples | | 403 | frame_skip_intensity | Indicates the | 0.0 | 404 | | intensity of the | | 405 | | frame skips | | 406 | since_last_frame_skip | Number of video | 0 | 407 | | frames since the | | 408 | | last skip | | 409 | consecutive_frame_skips | Number of | 0 | 410 | | consecutive frame | | 411 | | skips | | 412 | target_bitrate | Video target | TARGET_BITRATE_MIN | 413 | | bitrate [bps] | | 414 | target_bitrate_i | Video target | 1 | 415 | | bitrate inflection | | 416 | | point i.e. the | | 417 | | last known highest | | 418 | | target_bitrate | | 419 | | during fast start. | | 420 | | Used to limit | | 421 | | bitrate increase | | 422 | | close to the last | | 423 | | know congestion | | 424 | | point | | 425 | rate_transmit | Measured transmit | 0.0 | 426 | | bitrate [bps] | | 427 | rate_acked | Measured | 0.0 | 428 | | throughput based | | 429 | | on received | | 430 | | acknowledgements | | 431 | | [bps] | | 432 | s_rtt | Smoothed RTT [s], | 0.0 | 433 | | computed similar | | 434 | | to method depicted | | 435 | | in [RFC6298] | | 436 | rtp_queue_size | Size of RTP | 0 | 437 | | packets in queue | | 438 | | [bits] | | 439 | rtp_size | Size of the last | 0 | 440 | | transmitted RTP | | 441 | | packets [byte] | | 442 | frame_skip | Skip encoding of | false | 443 | | video frame if | | 444 | | true | | 445 +-------------------------+--------------------+--------------------+ 447 Table 2: State variables 449 +------------------+------------------------------------------------+ 450 | Variable | Explanation | 451 +------------------+------------------------------------------------+ 452 | owd | OWD = One way delay with base delay subtracted | 453 | | [s]. This is an estimate of the network | 454 | | queueing delay. | 455 | owd_fraction | OWD as a fraction of the OWD target | 456 | owd_norm | OWD normalized to OWD_TARGET_LO | 457 | owd_norm_mean | Average OWD norm over the last 100 samples | 458 | owd_norm_mean_sh | Average OWD norm over the last 20 samples | 459 | owd_norm_var | OWD norm variance over the last 100 samples | 460 | off_target | Relation between OWD and OWD target | 461 | scl_i | A general scalefactor that is applied to the | 462 | | CWND or target_bitrate increase | 463 | x_cwnd | Additional increase of CWND, used when | 464 | | send_wnd is computed | 465 | pace_bitrate | The allowed RTP packet transmission rate, used | 466 | | in the computation of t_pace [bps] | 467 | age_avg | Average RTP queue delay [s] | 468 | increment | Allowed target_bitrate increase | 469 | current_rate | Max of rate_transmit and rate_acked | 470 +------------------+------------------------------------------------+ 472 Table 3: Local temporary variables 474 4.1.2. Network congestion control 476 This section explains the network congestion control, it contains two 477 main functions 479 o Computation of congestion window at the sender: Gives an upper 480 limit to the number of bytes in flight i.e. how many bytes that 481 have been transmitted but not yet acknowledged. 483 o Transmission scheduling at the sender: RTP packets are transmitted 484 if allowed by the relation between the number of bytes in flight 485 and the congestion window. This is controlled by the send window. 487 Unlike TCP, SCReAM is not a byte oriented protocol, rather it is an 488 RTP packet oriented protocol. Thus it keeps a list of transmitted 489 RTP packets and their respective sending times (wall-clock time). 490 The feedback indicates the highest received RTP sequence number and a 491 timestamp (wall-clock time) when it was received. In addition, an 492 ACK list is included to make it possible to determine lost packets. 494 4.1.2.1. Congestion window update 496 The congestion window is computed from the one way (extra) delay 497 estimates (OWD) that are obtained from the send and received 498 timestamp of the RTP packets. LEDBAT [RFC6817] explains the details 499 of the computation of the OWD. An OWD sample is obtained for each 500 received acknowledgement. No smoothing of the OWD samples occur, 501 however some smoothing occurs anyway as the computation of the CWND 502 is in itself a low pass filter function. 504 SCReAM uses the terminology "Bytes in flight (bytes_in_flight)" which 505 is computed as the sum of the sizes of the RTP packets ranging from 506 the RTP packet most recently transmitted down to but not including 507 the acknowledged packet with the highest sequence number. As an 508 example: If RTP packet was sequence number SN with transmitted and 509 the last ACK indicated SN-5 as the highest received sequence number 510 then bytes in flight is computed as the sum of the size of RTP 511 packets with sequence number SN-4, SN-3, SN-2, SN-1 and SN. 513 CWND is updated differently depending on whether the congestion 514 control is in fast start or not and if a loss event is detected. A 515 Boolean variable in_fast_start indicates if the congestion is in fast 516 start state. 518 A loss event indicates one or more lost RTP packets within an RTT. 519 This is detected by means of inspection for holes in the sequence 520 number space in the acknowledgements with some margin for possible 521 packet reordering in the network. As an alternative, a timer for 522 loss detection similar to TCP RACK may be used. 524 Below is described the actions when an acknowledgement from the 525 receiver is received. 527 bytes_newly_acked is updated. 529 The OWD fraction and an average of it are computed as 531 owd_fraction = owd/owd_target 533 owd_fraction_avg = 0.9* owd_fraction_avg + 0.1* owd_fraction 535 The OWD fraction is sampled every 50ms and the last 20 samples are 536 stored in a vector (owd_fraction_hist). This vector is used in the 537 computation of an OWD trend that gives a value between 0.0 and 1.0 538 depending on how close to congestion it is. The OWD trend is 539 calculated as follows 541 Let R(owd_fraction_hist,K) be the autocorrelation function of 542 owd_fraction_hist at lag K. The 1st order prediction coefficient is 543 formulated as 545 a = R(owd_fraction_hist,1)/R(owd_fraction_hist,0) 547 The prediction coefficient a has positive values if OWD shows an 548 increasing trend, thus an indication of congestion is obtained before 549 the OWD target is reached. The prediction coefficient is further 550 multiplied with owd_fraction_avg to reduce sensitivity to increasing 551 OWD when OWD is very small. The OWD trend is thus computed as 553 owd_trend = max(0.0,min(1.0,a*owd_fraction_avg)) 555 The owd_trend is utilized in the media rate control and to determine 556 when to exit slow start. 558 An off target value is computed as 560 off_target = (owd_target - owd) / owd_target 562 A temporal variable is scl_i is computed as 564 scl_i = max(0.2, min(1.0, (abs(cwnd-cwnd_i)/cwnd_i*4)^2)) 566 scl_i is used to limit the CWND increase when close to the last known 567 max value, before congestion was last detected. 569 The congestion window update depends on whether a loss event has 570 occurred, and if the congestion control is if fast start or not. 572 ____________________________________________________________________ 574 On loss event: 576 If a loss event is detected then in_fast_start is set to false and 577 CWND is updated according to 579 cwnd_i = cwnd 581 cwnd = max(min_cwnd,cwnd*BETA) 583 otherwise the CWND update continues 584 ____________________________________________________________________ 586 in_fast_start = true: 588 in_fast_start is set to false and cwnd_i=cwnd if owd_trend >= 0.2 and 589 otherwise CWND is updated according to 591 cwnd = cwnd + bytes_newly_acked*scl_i 593 ____________________________________________________________________ 595 in_fast_start = false: 597 Values of off_target > 0.0 indicates that the congestion window can 598 be increased. This is done according to the equations below. 600 gain = GAIN*(1.0 + max(0.0, 1.0 - owd_trend/ 0.2)) 602 The equation above limits the gain when near congestion is detected 604 gain *= scl_i 606 This equation limits the gain when CWND is close to its last known 607 max value 609 cwnd += gain * off_target * bytes_newly_acked * mss / cwnd 611 Values of off_target <= 0.0 indicates congestion, CWND is then 612 updated according to the equation 614 cwnd += GAIN*off_target*bytes_newly_acked*mss/cwnd 616 The equations above are very similar to what is specified in 617 [RFC6817]. There are however a few differences. 619 o [RFC6817] specifies a constant GAIN, this specification however 620 limits the gain when CWND is increased dependent on near 621 congestion state and the relation to the last known max CWND 622 value. 624 o [RFC6817] specifies that the CWND increased is limited by an 625 additional function controlled by a constant ALLOWED_INCREASE. 626 This additional limitation is removed in this specification. 628 ____________________________________________________________________ 630 A number of final steps in the congestion window update procedure are 631 outlined below 632 ____________________________________________________________________ 634 Resume fast start: 636 Fast start can be resumed in order to speed up the bitrate increase 637 in case congestion abates. The condition to resume fast start 638 (in_fast_start = true) is that owd_trend is less than 0.2 for 1.0 639 seconds or more. 641 ____________________________________________________________________ 643 Competing flows compensation, adjustment of owd_target: 645 Competing flows compensation is needed to avoid that flows congestion 646 controlled by SCReAM are starved out by flows that are more 647 aggressive in their nature. The owd_target is adjusted according to 648 the owd_norm_mean_sh whenever owd_norm_var is below a given value. 649 The condition to update owd_target is fulfilled if owd_norm_var < 650 0.16 (indicating that the standard deviation is less than 0.4). 651 owd_target is then update as: 653 owd_target = min(OWD_TARGET_HI,max(OWD_TARGET_LO, owd_norm_mean_sh* 654 OWD_TARGET_LO*1.1)) 656 ____________________________________________________________________ 658 Final CWND adjustment step: 660 The congestion window is limited by the maximum number of bytes in 661 flight over the last 1.0 seconds according to 663 cwnd = min(cwnd, max_bytes_in_flight*MAX_BYTES_IN_FLIGHT_HEAD_ROOM) 665 This avoids possible over-estimation of the throughput after for 666 example, idle periods. 668 Finally cwnd is set to ensure that it is at least min_cwnd 670 cwnd = max(cwnd, MIN_CWND) 672 4.1.2.2. Transmission scheduling 674 The principle is to allow packet transmission of an RTP packet only 675 if the number of bytes in flight is less than the congestion window. 676 There are however two reasons why this strict rule will not work 677 optimally: 679 o Bitrate variations: The video frame size is always varying to a 680 larger or smaller extent, a strict rule as the one given above 681 will have the effect that the video bitrate have difficulties to 682 increase as the congestion window puts a too hard restriction on 683 the video frame size variation, this further can lead to 684 occasional queuing of RTP packets in the RTP packet queue that 685 will prevent bitrate increase because of the increased RTP queue 686 size. 688 o Reverse (feedback) path congestion: Especially in transport over 689 buffer-bloated networks, the one way delay in the reverse 690 direction may jump due to congestion. The effect of this is that 691 the acknowledgements are delayed with the result that the self- 692 clocking is temporarily halted, even though the forward path is 693 not congested. 695 Packets are transmitted at a pace given by the send window, computed 696 below 698 The send window is computed differently depending on OWD and its 699 relation to the OWD target. 701 o If owd > owd_target: 702 The send window is computed as 703 send_wnd = cwnd-bytes_in_flight 704 This enforces a strict rule that helps to prevent further queue 705 buildup. 707 o If owd <= owd_target: 708 A helper variable 709 x_cwnd=1.0+BYTES_IN_FLIGHT_SLACK*max(0.0, 710 min(1.0,1.0-owd_trend/0.5))/100.0 711 is computed. The send window is computed as 712 send_wnd = max(cwnd*x_cwnd, cwnd+mss)-bytes_in_flight 713 This gives a slack that reduces as congestion increases, 714 BYTES_IN_FLIGHT_SLACK is a maximum allowed slack in percent. A 715 large value increases the robustness to bitrate variations in the 716 source and congested feedback channel issues. The possible 717 drawback is increased delay or packet loss when forward path 718 congestion occur. 720 4.1.3. Video rate control 722 The video rate control is operated based on the size of the RTP 723 packet send queue and observed loss events. In addition, owd_trend 724 is also considered in the rate control, this to reduce the amount of 725 induced network jitter. 727 A variable target_bitrate is adjusted depending on the congestion 728 state. The target bitrate can vary between a minimum value 729 (target_bitrate_min) and a maximum value (target_bitrate_max). 731 For the overall bitrate adjustment, two network throughput estimates 732 are computed : 734 o rate_transmit: The measured transmit bitrate 736 o rate_acked: The ACKed bitrate, i.e. the volume of ACKed bits per 737 time unit. 739 Both estimates are updated every 200ms. 741 The current throughput current_rate is computed as the maximum value 742 of rate_transmit and rate_acked. The rationale behind the use of 743 rate_acked in addition to rate_transmit is that rate_transmit is 744 affected also by the amount of data that is available to transmit, 745 thus a lack of data to transmit can be seen as reduced throughput 746 that may itself cause an unnecessary rate reduction. To overcome 747 this shortcoming; rate_acked is used as well. This gives a more 748 stable throughput estimate. 750 The bitrate is updated at regular intervals, given by 751 RATE_ADJUST_INTERVAL and differently depending the fast start state 753 The rate change behavior depends on whether a loss event has 754 occurred, and if the congestion control is if fast start or not. 756 ____________________________________________________________________ 758 On loss event: 760 First of all the target_bitrate is updated if a new loss event was 761 indicated and the rate change procedure is exited. 763 target_bitrate_i = target_bitrate 765 target_bitrate = max(BETA_R* target_bitrate, TARGET_BITRATE_MIN) 767 If no loss event was indicated then the rate change procedure 768 continues. 770 ____________________________________________________________________ 772 in_fast_start = true: 774 An allowed increment is computed based on the congestion level and 775 the relation to target_bitrate_i 777 scl_i = (target_bitrate - target_bitrate_i)/ target_bitrate_i 779 increment = TARGET_BITRATE_MAX* RATE_ADJUST_INTERVAL/RAMP_UP_TIME* 780 (1.0- min(1.0, owd_trend/0.1)) 782 increment *= max(0.2, min(1.0, (scl_i*4)^2)) 784 target_bitrate += increment 786 target_bitrate is reduced further if congestion is detected. 788 target_bitrate *= (1.0- PRE_CONGESTION_GUARD*owd_trend) 790 target_bitrate = 791 min(TARGET_BITRATE_MAX,max(TARGET_BITRATE_MIN,target_bitrate)) 793 ____________________________________________________________________ 795 in_fast_start = false: 797 target_bitrate_i is updated to the current value of target_bitrate if 798 in_fast_start was true the last time the bitrate was updated. 800 A pre-congestion indicator is computed as 802 pre_congestion = min(1.0, max(0.0, owd_fraction_avg-0.3)/0.7) 804 pre_congestion += owd_trend 806 The target bitrate is computed as 808 target_bitrate=current_rate*(1.0- 809 PRE_CONGESTION_GUARD*pre_congestion)-TX_QUEUE_SIZE_FACTOR 810 *rtp_queue_size 812 target_bitrate = 813 min(TARGET_BITRATE_MAX,max(TARGET_BITRATE_MIN,target_bitrate)) 815 4.2. SCReAM Receiver 817 The SCReAM receiver is very simple in its implementation. The task 818 is to feedback acknowledgements of received packets. For that 819 purpose a set of state variables are needed, these are explained in 820 Table 4. 822 One set of state variables are maintained per stream. 824 +-----------------------------+-----------------------------+-------+ 825 | Variable | Explanation | Init | 826 | | | value | 827 +-----------------------------+-----------------------------+-------+ 828 | rx_timestamp | The wall clock timestamp | 0 | 829 | | when the latest RTP packet | | 830 | | was received | | 831 | highest_rtp_sequence_number | The highest received | 0 | 832 | | sequence number | | 833 | highest_rtp_timestamp | The highest received RTP | 0 | 834 | | timestamp, the RTP | | 835 | | timestamp is not needed by | | 836 | | the SCReAM algorithm but is | | 837 | | here used to resolve | | 838 | | possible RTP sequence | | 839 | | number wrap-around | | 840 | ack_vector | A 32 bit vector that | 0 | 841 | | indicates received RTP | | 842 | | packets with a sequence | | 843 | | number lower than | | 844 | | highest_rtp_sequence_number | | 845 | pending_feedback | Indicates that an RTP | false | 846 | | packet was received and | | 847 | | that an RTCP packet can be | | 848 | | generated when RTCP timing | | 849 | | rules permit | | 850 | last_transmit_t | Last time an RTCP packet | -1.0 | 851 | | was transmitted, this is | | 852 | | used to ensure that RTCP | | 853 | | feedback is generated | | 854 | | fairly for all streams. | | 855 +-----------------------------+-----------------------------+-------+ 857 Table 4: State variables 859 Upon reception of an RTP packet, the state variables in Table 4 860 should be updated and the RTCP processing function should be 861 notified. An RTCP packet is later generated based on the state 862 variables, how often this is done depends on the RTCP bandwidth. 864 5. Feedback Message 866 The feedback is over RTCP [RFC3550] and is based on [RFC4585]. It is 867 implemented as a transport layer feedback message (RTPFB), see 868 proposed example in Figure 2. The feedback control information part 869 (FCI) consists of the following elements. 871 o Timestamp: A timestamp value indicating when the last packet was 872 received which makes it possible to compute the one way (extra) 873 delay (OWD). 875 o The ACK list (Highest received sequence number + ACK vector): 876 Makes it possible to detect lost packets and determine the number 877 of bytes in flight. 879 o Source quench bit (Q): Makes it possible to request the sender to 880 reduce its congestion window. This is useful if WebRTC media is 881 received from many hosts and it becomes necessary to balance the 882 bitrates between the streams. 884 o ECE, ECN (Explicit Congestion Notification) echo: Makes it 885 possible to indicate if packets are ECN-CE (ECN Congestion 886 Experienced) marked. The use for the ECN echo bits is T.B.D. 888 o R : Reserved bits 890 o PT_RTP : Payload type number of RTP packet, along with the SSRC of 891 the media source this identifies the stream. 893 0 1 2 3 894 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 895 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 896 |V=2|P| FMT | PT | length | 897 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 898 | SSRC of packet sender | 899 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 900 | SSRC of media source | 901 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 902 | Timestamp (32bits) | 903 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 904 | Highest recv. seq. nr. (16b) | PT_RTP |Q|R|R| ECE | 905 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 906 | ACK vector (32b) | 907 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 909 Figure 2: Transport layer feedback message 911 To make the feedback as frequent as possible, the feedback packets 912 are transmitted as reduced size RTCP according to [RFC5506]. 914 The timestamp clock time is recommended to be set to a fixed value 915 such as 1000Hz, defined in this specification. The ACK vector is 916 here a bit vector that indicates the reception of the last 1+32 = 33 917 RTP packets. 919 Section 4 describes the main algorithm details and how the feedback 920 is used. 922 6. Additional features 924 This section describes additional features. They are not required 925 for the basic functionality of SCReAM but can improve performance in 926 certain scenarios and topologies. 928 6.1. Packet pacing 930 Packet pacing is used in order to mitigate coalescing i.e. that 931 packets are transmitted in bursts. 933 Packet pacing is enforced when owd_fraction_avg is greater than 0.1. 934 The time interval between consecutive packet transmissions is then 935 enforced to equal or higher than t_pace where t_pace is given by the 936 equations below. 938 pace_bitrate = max (50000, cwnd* 8 / s_rtt) 940 t_pace = rtp_size * 8 / pace_bitrate 942 rtp_size is the size of the last transmitted RTP packet 944 6.2. Frame skipping 946 Frame skipping is a feature that makes it possible to reduce the size 947 of the RTP queue in the cases that e.g. the channel throughput drops 948 dramatically or even goes below the lowest possible video coder rate. 949 Frame skipping is optional to implement as it can sometimes be 950 difficult to realize e.g. due to lack of API function to support 951 this. 953 Frame skipping is controlled by a flag frame_skip which, if set to 1 954 dictates that the video coder should skip the next video frame. The 955 frame skipping intensity at the current time instant is computed 956 according to the steps below 957 The queuing delay is sampled every frame period and the last 20 958 samples are stored in a vector age_vec 960 An average queuing delay is computed as a weighted sum over the 961 samples in age_vec. age_avg at the current time instant is computed 962 as 964 age_avg(n) = SUM age_vec(n-k)*w(k) k = [0..20[ 966 w(n) are weight factors arranged to give the most recent samples a 967 higher weight. 969 The change in age_avg is computed as 971 age_d = age_avg(n) - age_avg(n-1) 973 The frame skipping intensity at the current time instant n is 974 computed as 976 o If age_d > 0 and age_avg > 2*FRAME_PERIOD: 977 frame_skip_intensity = min(1.0, (age_vec(n)-2*FRAME_PERIOD)/(4* 978 FRAME_PERIOD) 980 o Otherwise frame skip intensity is set to zero 982 The skip_frame flag is set depending on three variables 984 o frame_skip_intensity 986 o since_last_frame_skip, i.e the number of consecutive frames 987 without frame skipping 989 o consecutive_frame_skips, i.e the number of consecutive frame skips 991 The flag skip_frame is set to 1 if any of the conditions below is 992 met, otherwise it is set to 0. 994 o age_vec(n) > 0.2 && consecutive_frame_skips < 5 996 o frame_skip_intensity < 0.5 && since_last_frame_skip >= 1.0/ 997 frame_skip_intensity 999 o frame_skip_intensity >= 0.5 && consecutive_frame_skips < 1000 (frame_skip_intensity -0.5)*10 1002 The arrangement makes sure that no more than 4 frames are skipped in 1003 sequence, the rationale is to ensure that the input to the video 1004 encoder does not change to much, something that may give poor 1005 prediction gain. 1007 6.3. Q-bit semantics (source quench) 1009 The Q bit in the feedback is set by a receiver to signal that the 1010 sender should reduce the bitrate. The sender will in response to 1011 this reduce the congestion window with the consequence that the video 1012 bitrate decreases. A typical use case for source quench is when a 1013 receiver receives streams from sources located at different hosts and 1014 they all share a common bottleneck, typically it is difficult to 1015 apply any rate distribution signaling between the sending hosts. The 1016 solution is then that the receiver sets the Q bit in the feedback to 1017 the sender that should reduce its rate, if the streams share a common 1018 bottleneck then the released bandwidth due to the reduction of the 1019 congestion window for the flow that had the Q bit set in the feedback 1020 will be grabbed by the other flows that did not have the Q bit set. 1021 This is ensured by the opportunistic behavior of SCReAM's congestion 1022 control. The source quench will have no or little effect if the 1023 flows do not share the same bottleneck. 1025 The reduction in congestion window is proportional to the amount of 1026 SCReAM RTCP feedback with the Q bit set, the below steps outline how 1027 the sender should react to RTCP feedback with the Q bit set. The 1028 reduction is done once per RTT. Let : 1030 o n = Number of received RTCP feedback messages in one RTT 1032 o n_q = Number of received RTCP feedback messages in one RTT, with Q 1033 bit set. 1035 The new congestion window is then expressed as: 1037 cwnd = max(MIN_CWND, cwnd*(1.0-0.5* n_q /n)) 1039 Note that CWND is adjusted at most once per RTT. Furthermore The 1040 CWND increase should be inhibited for one RTT if CWND has been 1041 decreased as a result of Q bits set in the feedback. 1043 The required intensity of the Q-bit set in the feedback in order to 1044 achieve a given rate distribution depends on many factors such as 1045 RTT, video source material etc. The receiver thus need to monitor 1046 the change in the received video bitrate on the different streams and 1047 adjust the intensity of the Q-bit accordingly. 1049 7. Discussion 1051 This section covers a few open discussion points 1053 o RTCP feedback overhead: SCReAM benefits from a relatively frequent 1054 feedback. Experiments have shown that a feedback rate roughly 1055 equal to the frame rate gives a stable self-clocking and 1056 robustness against loss of feedback. With a maximum bitrate of 1057 1500kbps the RTCP feedback overhead is in the range 10-15kbps with 1058 reduced size RTCP, including IP and UDP framing, in other words 1059 the RTCP overhead is quite modest and should not pose a problem in 1060 the general case. Other solutions may be required in highly 1061 asymmetrical link capacity cases. Worth notice is that SCReAM can 1062 work with as low feedback rates as once every 200ms, this however 1063 comes with a higher sensitivity to loss of feedback and also a 1064 potential reduction in throughput. 1066 o AVPF mode: The RTCP feedback is based on AVPF regular mode. The 1067 SCReAM feedback is transmitted as reduced size RTCP so save 1068 overhead, it is however required to transmit full compound RTCP at 1069 regular intervals, this interval can be controlled by trr-int 1070 depicted in [RFC4585]. 1072 o BETA, CWND scale factor due to loss: The BETA value is recommended 1073 to be higher than 0.5. The reason behind this is that congestion 1074 control for multimedia has to deal with a source that is rate 1075 limited. A file transfer has "unlimited" source bitrate in 1076 comparison. The outcome is that SCReAM must be a little more 1077 aggressive than a file transfer in order to not be out competed. 1079 8. Conclusion 1081 This memo describes a congestion control algorithm for RMCAT that it 1082 is particularly good at handling the quickly changing condition in 1083 wireless network such as LTE. The solution conforms to the packet 1084 conservation principle and leverages on novel congestion control 1085 algorithms and recent TCP research, together with media bitrate 1086 determined by sender queuing delay and given delay thresholds. The 1087 solution has shown potential to meet the goals of high link 1088 utilization and prompt reaction to congestion. The solution is 1089 realized with a new RFC4585 transport layer feedback message. 1091 9. Open issues 1093 A list of open issues. 1095 o Describe how clock drift compensation is done 1096 o Determine format and use of ECN echo field 1098 o Describe how FEC overhead is accounted for in target_bitrate 1099 computation 1101 o Describe how current_rate is distributed between two or more RTP 1102 streams 1104 o Investigate the impact of more sparse RTCP feedback, for instance 1105 once per RTT 1107 10. Acknowledgements 1109 We would like to thank the following persons for their comments, 1110 questions and support during the work that led to this memo: Markus 1111 Andersson, Bo Burman, Tomas Frankkila, Frederic Gabin, Laurits Hamm, 1112 Hans Hannu, Nikolas Hermanns, Stefan Haekansson, Erlendur Karlsson, 1113 Daniel Lindstroem, Mats Nordberg, Jonathan Samuelsson, Rickard 1114 Sjoeberg, Robert Swain, Magnus Westerlund. 1116 11. IANA Considerations 1118 A new RFC4585 transport layer feedback message needs to be 1119 standardized. 1121 12. Security Considerations 1123 The feedback can be vulnerable to attacks similar to those that can 1124 affect TCP. It is therefore recommended that the RTCP feedback is at 1125 least integrity protected. 1127 13. Change history 1129 A list of changes: 1131 o -03 to -04 : Extensive changes due to review comments, code 1132 somewhat modified, frame skipping made optional 1134 o -02 to -03 : Added algorithm description with equations, removed 1135 pseudo code and simulation results 1137 o -01 to -02 : Updated GCC simulation results 1139 o -00 to -01 : Fixed a few bugs in example code 1141 14. References 1143 14.1. Normative References 1145 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1146 Requirement Levels", BCP 14, RFC 2119, March 1997. 1148 [RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and V. 1149 Jacobson, "RTP: A Transport Protocol for Real-Time 1150 Applications", STD 64, RFC 3550, July 2003. 1152 [RFC4585] Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey, 1153 "Extended RTP Profile for Real-time Transport Control 1154 Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585, July 1155 2006. 1157 [RFC5506] Johansson, I. and M. Westerlund, "Support for Reduced-Size 1158 Real-Time Transport Control Protocol (RTCP): Opportunities 1159 and Consequences", RFC 5506, April 2009. 1161 [RFC6298] Paxson, V., Allman, M., Chu, J., and M. Sargent, 1162 "Computing TCP's Retransmission Timer", RFC 6298, June 1163 2011. 1165 [RFC6817] Shalunov, S., Hazel, G., Iyengar, J., and M. Kuehlewind, 1166 "Low Extra Delay Background Transport (LEDBAT)", RFC 6817, 1167 December 2012. 1169 14.2. Informative References 1171 [FACK] "Forward Acknowledgement: Refining TCP Congestion 1172 Control", 2006. 1174 [I-D.draft-sarker-rmcat-cellular-eval-test-cases] 1175 Sarker, Z., "Evaluation Test Cases for Interactive Real- 1176 Time Media over Cellular Networks", 1177 . 1180 [I-D.ietf-rmcat-app-interaction] 1181 Zanaty, M., Singh, V., Nandakumar, S., and Z. Sarker, "RTP 1182 Application Interaction with Congestion Control", draft- 1183 ietf-rmcat-app-interaction-01 (work in progress), October 1184 2014. 1186 [I-D.ietf-tcpm-newcwv] 1187 Fairhurst, G., Sathiaseelan, A., and R. Secchi, "Updating 1188 TCP to support Rate-Limited Traffic", draft-ietf-tcpm- 1189 newcwv-07 (work in progress), September 2014. 1191 [QoS-3GPP] 1192 TS 23.203, 3GPP., "Policy and charging control 1193 architecture", June 2011, . 1196 [TFWC] University College London, "Fairer TCP-Friendly Congestion 1197 Control Protocol for Multimedia Streaming", December 2007, 1198 . 1201 Authors' Addresses 1203 Ingemar Johansson 1204 Ericsson AB 1205 Laboratoriegraend 11 1206 Luleae 977 53 1207 Sweden 1209 Phone: +46 730783289 1210 Email: ingemar.s.johansson@ericsson.com 1212 Zaheduzzaman Sarker 1213 Ericsson AB 1214 Laboratoriegraend 11 1215 Luleae 977 53 1216 Sweden 1218 Phone: +46 761153743 1219 Email: zaheduzzaman.sarker@ericsson.com