idnits 2.17.1 draft-ietf-rtcweb-security-09.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 : ---------------------------------------------------------------------------- ** The document seems to lack an IANA Considerations section. (See Section 2.2 of https://www.ietf.org/id-info/checklist for how to handle the case when there are no actions for IANA.) Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == The document seems to contain a disclaimer for pre-RFC5378 work, but was first submitted on or after 10 November 2008. The disclaimer is usually necessary only for documents that revise or obsolete older RFCs, and that take significant amounts of text from those RFCs. If you can contact all authors of the source material and they are willing to grant the BCP78 rights to the IETF Trust, you can and should remove the disclaimer. Otherwise, the disclaimer is needed and you can ignore this comment. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (October 29, 2017) is 2369 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Unused Reference: 'I-D.kaufman-rtcweb-security-ui' is defined on line 1044, but no explicit reference was found in the text == Unused Reference: 'RFC2818' is defined on line 1055, but no explicit reference was found in the text == Outdated reference: A later version (-12) exists of draft-ietf-rtcweb-ip-handling-04 == Outdated reference: A later version (-19) exists of draft-ietf-rtcweb-overview-18 == Outdated reference: A later version (-20) exists of draft-ietf-rtcweb-security-arch-12 -- Obsolete informational reference (is this intentional?): RFC 2818 (Obsoleted by RFC 9110) -- Obsolete informational reference (is this intentional?): RFC 4347 (Obsoleted by RFC 6347) -- Obsolete informational reference (is this intentional?): RFC 5245 (Obsoleted by RFC 8445, RFC 8839) -- Obsolete informational reference (is this intentional?): RFC 6222 (Obsoleted by RFC 7022) Summary: 1 error (**), 0 flaws (~~), 7 warnings (==), 5 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 RTC-Web E. Rescorla 3 Internet-Draft RTFM, Inc. 4 Intended status: Standards Track October 29, 2017 5 Expires: May 2, 2018 7 Security Considerations for WebRTC 8 draft-ietf-rtcweb-security-09 10 Abstract 12 WebRTC is a protocol suite for use with real-time applications that 13 can be deployed in browsers - "real time communication on the Web". 14 This document defines the WebRTC threat model and analyzes the 15 security threats of WebRTC in that model. 17 Status of This Memo 19 This Internet-Draft is submitted in full conformance with the 20 provisions of BCP 78 and BCP 79. 22 Internet-Drafts are working documents of the Internet Engineering 23 Task Force (IETF). Note that other groups may also distribute 24 working documents as Internet-Drafts. The list of current Internet- 25 Drafts is at http://datatracker.ietf.org/drafts/current/. 27 Internet-Drafts are draft documents valid for a maximum of six months 28 and may be updated, replaced, or obsoleted by other documents at any 29 time. It is inappropriate to use Internet-Drafts as reference 30 material or to cite them other than as "work in progress." 32 This Internet-Draft will expire on May 2, 2018. 34 Copyright Notice 36 Copyright (c) 2017 IETF Trust and the persons identified as the 37 document authors. All rights reserved. 39 This document is subject to BCP 78 and the IETF Trust's Legal 40 Provisions Relating to IETF Documents 41 (http://trustee.ietf.org/license-info) in effect on the date of 42 publication of this document. Please review these documents 43 carefully, as they describe your rights and restrictions with respect 44 to this document. Code Components extracted from this document must 45 include Simplified BSD License text as described in Section 4.e of 46 the Trust Legal Provisions and are provided without warranty as 47 described in the Simplified BSD License. 49 This document may contain material from IETF Documents or IETF 50 Contributions published or made publicly available before November 51 10, 2008. The person(s) controlling the copyright in some of this 52 material may not have granted the IETF Trust the right to allow 53 modifications of such material outside the IETF Standards Process. 54 Without obtaining an adequate license from the person(s) controlling 55 the copyright in such materials, this document may not be modified 56 outside the IETF Standards Process, and derivative works of it may 57 not be created outside the IETF Standards Process, except to format 58 it for publication as an RFC or to translate it into languages other 59 than English. 61 Table of Contents 63 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 64 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4 65 3. The Browser Threat Model . . . . . . . . . . . . . . . . . . 4 66 3.1. Access to Local Resources . . . . . . . . . . . . . . . . 5 67 3.2. Same Origin Policy . . . . . . . . . . . . . . . . . . . 5 68 3.3. Bypassing SOP: CORS, WebSockets, and consent to 69 communicate . . . . . . . . . . . . . . . . . . . . . . . 6 70 4. Security for WebRTC Applications . . . . . . . . . . . . . . 7 71 4.1. Access to Local Devices . . . . . . . . . . . . . . . . . 7 72 4.1.1. Threats from Screen Sharing . . . . . . . . . . . . . 8 73 4.1.2. Calling Scenarios and User Expectations . . . . . . . 8 74 4.1.2.1. Dedicated Calling Services . . . . . . . . . . . 8 75 4.1.2.2. Calling the Site You're On . . . . . . . . . . . 9 76 4.1.3. Origin-Based Security . . . . . . . . . . . . . . . . 9 77 4.1.4. Security Properties of the Calling Page . . . . . . . 11 78 4.2. Communications Consent Verification . . . . . . . . . . . 12 79 4.2.1. ICE . . . . . . . . . . . . . . . . . . . . . . . . . 12 80 4.2.2. Masking . . . . . . . . . . . . . . . . . . . . . . . 13 81 4.2.3. Backward Compatibility . . . . . . . . . . . . . . . 13 82 4.2.4. IP Location Privacy . . . . . . . . . . . . . . . . . 15 83 4.3. Communications Security . . . . . . . . . . . . . . . . . 15 84 4.3.1. Protecting Against Retrospective Compromise . . . . . 16 85 4.3.2. Protecting Against During-Call Attack . . . . . . . . 17 86 4.3.2.1. Key Continuity . . . . . . . . . . . . . . . . . 17 87 4.3.2.2. Short Authentication Strings . . . . . . . . . . 18 88 4.3.2.3. Third Party Identity . . . . . . . . . . . . . . 19 89 4.3.2.4. Page Access to Media . . . . . . . . . . . . . . 19 90 4.3.3. Malicious Peers . . . . . . . . . . . . . . . . . . . 20 91 4.4. Privacy Considerations . . . . . . . . . . . . . . . . . 20 92 4.4.1. Correlation of Anonymous Calls . . . . . . . . . . . 20 93 4.4.2. Browser Fingerprinting . . . . . . . . . . . . . . . 20 94 5. Security Considerations . . . . . . . . . . . . . . . . . . . 21 95 6. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 21 96 7. Changes Since -04 . . . . . . . . . . . . . . . . . . . . . . 21 97 8. References . . . . . . . . . . . . . . . . . . . . . . . . . 21 98 8.1. Normative References . . . . . . . . . . . . . . . . . . 21 99 8.2. Informative References . . . . . . . . . . . . . . . . . 21 100 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . 24 102 1. Introduction 104 The Real-Time Communications on the Web (RTCWEB) working group is 105 tasked with standardizing protocols for real-time communications 106 between Web browsers, generally called "WebRTC" 107 [I-D.ietf-rtcweb-overview]. The major use cases for WebRTC 108 technology are real-time audio and/or video calls, Web conferencing, 109 and direct data transfer. Unlike most conventional real-time 110 systems, (e.g., SIP-based[RFC3261] soft phones) WebRTC communications 111 are directly controlled by some Web server. A simple case is shown 112 below. 114 +----------------+ 115 | | 116 | Web Server | 117 | | 118 +----------------+ 119 ^ ^ 120 / \ 121 HTTP / \ HTTP 122 or / \ or 123 WebSockets / \ WebSockets 124 v v 125 JS API JS API 126 +-----------+ +-----------+ 127 | | Media | | 128 | Browser |<---------->| Browser | 129 | | | | 130 +-----------+ +-----------+ 132 Figure 1: A simple WebRTC system 134 In the system shown in Figure 1, Alice and Bob both have WebRTC 135 enabled browsers and they visit some Web server which operates a 136 calling service. Each of their browsers exposes standardized 137 JavaScript calling APIs (implementated as browser built-ins) which 138 are used by the Web server to set up a call between Alice and Bob. 139 The Web server also serves as the signaling channel to transport 140 control messages between the browsers. While this system is 141 topologically similar to a conventional SIP-based system (with the 142 Web server acting as the signaling service and browsers acting as 143 softphones), control has moved to the central Web server; the browser 144 simply provides API points that are used by the calling service. As 145 with any Web application, the Web server can move logic between the 146 server and JavaScript in the browser, but regardless of where the 147 code is executing, it is ultimately under control of the server. 149 It should be immediately apparent that this type of system poses new 150 security challenges beyond those of a conventional VoIP system. In 151 particular, it needs to contend with malicious calling services. For 152 example, if the calling service can cause the browser to make a call 153 at any time to any callee of its choice, then this facility can be 154 used to bug a user's computer without their knowledge, simply by 155 placing a call to some recording service. More subtly, if the 156 exposed APIs allow the server to instruct the browser to send 157 arbitrary content, then they can be used to bypass firewalls or mount 158 denial of service attacks. Any successful system will need to be 159 resistant to this and other attacks. 161 A companion document [I-D.ietf-rtcweb-security-arch] describes a 162 security architecture intended to address the issues raised in this 163 document. 165 2. Terminology 167 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 168 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 169 document are to be interpreted as described in RFC 2119 [RFC2119]. 171 3. The Browser Threat Model 173 The security requirements for WebRTC follow directly from the 174 requirement that the browser's job is to protect the user. Huang et 175 al. [huang-w2sp] summarize the core browser security guarantee as: 177 Users can safely visit arbitrary web sites and execute scripts 178 provided by those sites. 180 It is important to realize that this includes sites hosting arbitrary 181 malicious scripts. The motivation for this requirement is simple: it 182 is trivial for attackers to divert users to sites of their choice. 183 For instance, an attacker can purchase display advertisements which 184 direct the user (either automatically or via user clicking) to their 185 site, at which point the browser will execute the attacker's scripts. 186 Thus, it is important that it be safe to view arbitrarily malicious 187 pages. Of course, browsers inevitably have bugs which cause them to 188 fall short of this goal, but any new WebRTC functionality must be 189 designed with the intent to meet this standard. The remainder of 190 this section provides more background on the existing Web security 191 model. 193 In this model, then, the browser acts as a TRUSTED COMPUTING BASE 194 (TCB) both from the user's perspective and to some extent from the 195 server's. While HTML and JavaScript (JS) provided by the server can 196 cause the browser to execute a variety of actions, those scripts 197 operate in a sandbox that isolates them both from the user's computer 198 and from each other, as detailed below. 200 Conventionally, we refer to either WEB ATTACKERS, who are able to 201 induce you to visit their sites but do not control the network, and 202 NETWORK ATTACKERS, who are able to control your network. Network 203 attackers correspond to the [RFC3552] "Internet Threat Model". Note 204 that for non-HTTPS traffic, a network attacker is also a Web 205 attacker, since it can inject traffic as if it were any non-HTTPS Web 206 site. Thus, when analyzing HTTP connections, we must assume that 207 traffic is going to the attacker. 209 3.1. Access to Local Resources 211 While the browser has access to local resources such as keying 212 material, files, the camera and the microphone, it strictly limits or 213 forbids web servers from accessing those same resources. For 214 instance, while it is possible to produce an HTML form which will 215 allow file upload, a script cannot do so without user consent and in 216 fact cannot even suggest a specific file (e.g., /etc/passwd); the 217 user must explicitly select the file and consent to its upload. 218 [Note: in many cases browsers are explicitly designed to avoid 219 dialogs with the semantics of "click here to screw yourself", as 220 extensive research shows that users are prone to consent under such 221 circumstances.] 223 Similarly, while Flash programs (SWFs) [SWF] can access the camera 224 and microphone, they explicitly require that the user consent to that 225 access. In addition, some resources simply cannot be accessed from 226 the browser at all. For instance, there is no real way to run 227 specific executables directly from a script (though the user can of 228 course be induced to download executable files and run them). 230 3.2. Same Origin Policy 232 Many other resources are accessible but isolated. For instance, 233 while scripts are allowed to make HTTP requests via the 234 XMLHttpRequest() API those requests are not allowed to be made to any 235 server, but rather solely to the same ORIGIN from whence the script 236 came [RFC6454] (although CORS [CORS] and WebSockets [RFC6455] provide 237 a escape hatch from this restriction, as described below.) This SAME 238 ORIGIN POLICY (SOP) prevents server A from mounting attacks on server 239 B via the user's browser, which protects both the user (e.g., from 240 misuse of his credentials) and the server B (e.g., from DoS attack). 242 More generally, SOP forces scripts from each site to run in their 243 own, isolated, sandboxes. While there are techniques to allow them 244 to interact, those interactions generally must be mutually consensual 245 (by each site) and are limited to certain channels. For instance, 246 multiple pages/browser panes from the same origin can read each 247 other's JS variables, but pages from the different origins--or even 248 iframes from different origins on the same page--cannot. 250 3.3. Bypassing SOP: CORS, WebSockets, and consent to communicate 252 While SOP serves an important security function, it also makes it 253 inconvenient to write certain classes of applications. In 254 particular, mash-ups, in which a script from origin A uses resources 255 from origin B, can only be achieved via a certain amount of hackery. 256 The W3C Cross-Origin Resource Sharing (CORS) spec [CORS] is a 257 response to this demand. In CORS, when a script from origin A 258 executes what would otherwise be a forbidden cross-origin request, 259 the browser instead contacts the target server to determine whether 260 it is willing to allow cross-origin requests from A. If it is so 261 willing, the browser then allows the request. This consent 262 verification process is designed to safely allow cross-origin 263 requests. 265 While CORS is designed to allow cross-origin HTTP requests, 266 WebSockets [RFC6455] allows cross-origin establishment of transparent 267 channels. Once a WebSockets connection has been established from a 268 script to a site, the script can exchange any traffic it likes 269 without being required to frame it as a series of HTTP request/ 270 response transactions. As with CORS, a WebSockets transaction starts 271 with a consent verification stage to avoid allowing scripts to simply 272 send arbitrary data to another origin. 274 While consent verification is conceptually simple--just do a 275 handshake before you start exchanging the real data--experience has 276 shown that designing a correct consent verification system is 277 difficult. In particular, Huang et al. [huang-w2sp] have shown 278 vulnerabilities in the existing Java and Flash consent verification 279 techniques and in a simplified version of the WebSockets handshake. 280 In particular, it is important to be wary of CROSS-PROTOCOL attacks 281 in which the attacking script generates traffic which is acceptable 282 to some non-Web protocol state machine. In order to resist this form 283 of attack, WebSockets incorporates a masking technique intended to 284 randomize the bits on the wire, thus making it more difficult to 285 generate traffic which resembles a given protocol. 287 4. Security for WebRTC Applications 289 4.1. Access to Local Devices 291 As discussed in Section 1, allowing arbitrary sites to initiate calls 292 violates the core Web security guarantee; without some access 293 restrictions on local devices, any malicious site could simply bug a 294 user. At minimum, then, it MUST NOT be possible for arbitrary sites 295 to initiate calls to arbitrary locations without user consent. This 296 immediately raises the question, however, of what should be the scope 297 of user consent. 299 In order for the user to make an intelligent decision about whether 300 to allow a call (and hence his camera and microphone input to be 301 routed somewhere), he must understand either who is requesting 302 access, where the media is going, or both. As detailed below, there 303 are two basic conceptual models: 305 1. You are sending your media to entity A because you want to talk 306 to Entity A (e.g., your mother). 308 2. Entity A (e.g., a calling service) asks to access the user's 309 devices with the assurance that it will transfer the media to 310 entity B (e.g., your mother) 312 In either case, identity is at the heart of any consent decision. 313 Moreover, the identity of the party the browser is connecting to is 314 all that the browser can meaningfully enforce; if you are calling A, 315 A can simply forward the media to C. Similarly, if you authorize A 316 to place a call to B, A can call C instead. In either case, all the 317 browser is able to do is verify and check authorization for whoever 318 is controlling where the media goes. The target of the media can of 319 course advertise a security/privacy policy, but this is not something 320 that the browser can enforce. Even so, there are a variety of 321 different consent scenarios that motivate different technical consent 322 mechanisms. We discuss these mechanisms in the sections below. 324 It's important to understand that consent to access local devices is 325 largely orthogonal to consent to transmit various kinds of data over 326 the network (see Section 4.2). Consent for device access is largely 327 a matter of protecting the user's privacy from malicious sites. By 328 contrast, consent to send network traffic is about preventing the 329 user's browser from being used to attack its local network. Thus, we 330 need to ensure communications consent even if the site is not able to 331 access the camera and microphone at all (hence WebSockets's consent 332 mechanism) and similarly we need to be concerned with the site 333 accessing the user's camera and microphone even if the data is to be 334 sent back to the site via conventional HTTP-based network mechanisms 335 such as HTTP POST. 337 4.1.1. Threats from Screen Sharing 339 In addition to camera and microphone access, there has been demand 340 for screen and/or application sharing functionality. Unfortunately, 341 the security implications of this functionality are much harder for 342 users to intuitively analyze than for camera and microphone access. 343 (See http://lists.w3.org/Archives/Public/public- 344 webrtc/2013Mar/0024.html for a full analysis.) 346 The most obvious threats are simply those of "oversharing". I.e., 347 the user may believe they are sharing a window when in fact they are 348 sharing an application, or may forget they are sharing their whole 349 screen, icons, notifications, and all. This is already an issue with 350 existing screen sharing technologies and is made somewhat worse if a 351 partially trusted site is responsible for asking for the resource to 352 be shared rather than having the user propose it. 354 A less obvious threat involves the impact of screen sharing on the 355 Web security model. A key part of the Same Origin Policy is that 356 HTML or JS from site A can reference content from site B and cause 357 the browser to load it, but (unless explicitly permitted) cannot see 358 the result. However, if a web application from a site is screen 359 sharing the browser, then this violates that invariant, with serious 360 security consequences. For example, an attacker site might request 361 screen sharing and then briefly open up a new Window to the user's 362 bank or webmail account, using screen sharing to read the resulting 363 displayed content. A more sophisticated attack would be open up a 364 source view window to a site and use the screen sharing result to 365 view anti cross-site request forgery tokens. 367 These threats suggest that screen/application sharing might need a 368 higher level of user consent than access to the camera or microphone. 370 4.1.2. Calling Scenarios and User Expectations 372 While a large number of possible calling scenarios are possible, the 373 scenarios discussed in this section illustrate many of the 374 difficulties of identifying the relevant scope of consent. 376 4.1.2.1. Dedicated Calling Services 378 The first scenario we consider is a dedicated calling service. In 379 this case, the user has a relationship with a calling site and 380 repeatedly makes calls on it. It is likely that rather than having 381 to give permission for each call that the user will want to give the 382 calling service long-term access to the camera and microphone. This 383 is a natural fit for a long-term consent mechanism (e.g., installing 384 an app store "application" to indicate permission for the calling 385 service.) A variant of the dedicated calling service is a gaming 386 site (e.g., a poker site) which hosts a dedicated calling service to 387 allow players to call each other. 389 With any kind of service where the user may use the same service to 390 talk to many different people, there is a question about whether the 391 user can know who they are talking to. If I grant permission to 392 calling service A to make calls on my behalf, then I am implicitly 393 granting it permission to bug my computer whenever it wants. This 394 suggests another consent model in which a site is authorized to make 395 calls but only to certain target entities (identified via media-plane 396 cryptographic mechanisms as described in Section 4.3.2 and especially 397 Section 4.3.2.3.) Note that the question of consent here is related 398 to but distinct from the question of peer identity: I might be 399 willing to allow a calling site to in general initiate calls on my 400 behalf but still have some calls via that site where I can be sure 401 that the site is not listening in. 403 4.1.2.2. Calling the Site You're On 405 Another simple scenario is calling the site you're actually visiting. 406 The paradigmatic case here is the "click here to talk to a 407 representative" windows that appear on many shopping sites. In this 408 case, the user's expectation is that they are calling the site 409 they're actually visiting. However, it is unlikely that they want to 410 provide a general consent to such a site; just because I want some 411 information on a car doesn't mean that I want the car manufacturer to 412 be able to activate my microphone whenever they please. Thus, this 413 suggests the need for a second consent mechanism where I only grant 414 consent for the duration of a given call. As described in 415 Section 3.1, great care must be taken in the design of this interface 416 to avoid the users just clicking through. Note also that the user 417 interface chrome must clearly display elements showing that the call 418 is continuing in order to avoid attacks where the calling site just 419 leaves it up indefinitely but shows a Web UI that implies otherwise. 421 4.1.3. Origin-Based Security 423 Now that we have seen another use case, we can start to reason about 424 the security requirements. 426 As discussed in Section 3.2, the basic unit of Web sandboxing is the 427 origin, and so it is natural to scope consent to origin. 428 Specifically, a script from origin A MUST only be allowed to initiate 429 communications (and hence to access camera and microphone) if the 430 user has specifically authorized access for that origin. It is of 431 course technically possible to have coarser-scoped permissions, but 432 because the Web model is scoped to origin, this creates a difficult 433 mismatch. 435 Arguably, origin is not fine-grained enough. Consider the situation 436 where Alice visits a site and authorizes it to make a single call. 437 If consent is expressed solely in terms of origin, then at any future 438 visit to that site (including one induced via mash-up or ad network), 439 the site can bug Alice's computer, use the computer to place bogus 440 calls, etc. While in principle Alice could grant and then revoke the 441 privilege, in practice privileges accumulate; if we are concerned 442 about this attack, something else is needed. There are a number of 443 potential countermeasures to this sort of issue. 445 Individual Consent 447 Ask the user for permission for each call. 449 Callee-oriented Consent 451 Only allow calls to a given user. 453 Cryptographic Consent 455 Only allow calls to a given set of peer keying material or to a 456 cryptographically established identity. 458 Unfortunately, none of these approaches is satisfactory for all 459 cases. As discussed above, individual consent puts the user's 460 approval in the UI flow for every call. Not only does this quickly 461 become annoying but it can train the user to simply click "OK", at 462 which point the consent becomes useless. Thus, while it may be 463 necessary to have individual consent in some case, this is not a 464 suitable solution for (for instance) the calling service case. Where 465 necessary, in-flow user interfaces must be carefully designed to 466 avoid the risk of the user blindly clicking through. 468 The other two options are designed to restrict calls to a given 469 target. Callee-oriented consent provided by the calling site not 470 work well because a malicious site can claim that the user is calling 471 any user of his choice. One fix for this is to tie calls to a 472 cryptographically established identity. While not suitable for all 473 cases, this approach may be useful for some. If we consider the case 474 of advertising, it's not particularly convenient to require the 475 advertiser to instantiate an iframe on the hosting site just to get 476 permission; a more convenient approach is to cryptographically tie 477 the advertiser's certificate to the communication directly. We're 478 still tying permissions to origin here, but to the media origin (and- 479 or destination) rather than to the Web origin. 480 [I-D.ietf-rtcweb-security-arch] describes mechanisms which facilitate 481 this sort of consent. 483 Another case where media-level cryptographic identity makes sense is 484 when a user really does not trust the calling site. For instance, I 485 might be worried that the calling service will attempt to bug my 486 computer, but I also want to be able to conveniently call my friends. 487 If consent is tied to particular communications endpoints, then my 488 risk is limited. Naturally, it is somewhat challenging to design UI 489 primitives which express this sort of policy. The problem becomes 490 even more challenging in multi-user calling cases. 492 4.1.4. Security Properties of the Calling Page 494 Origin-based security is intended to secure against web attackers. 495 However, we must also consider the case of network attackers. 496 Consider the case where I have granted permission to a calling 497 service by an origin that has the HTTP scheme, e.g., http://calling- 498 service.example.com. If I ever use my computer on an unsecured 499 network (e.g., a hotspot or if my own home wireless network is 500 insecure), and browse any HTTP site, then an attacker can bug my 501 computer. The attack proceeds like this: 503 1. I connect to http://anything.example.org/. Note that this site is 504 unaffiliated with the calling service. 506 2. The attacker modifies my HTTP connection to inject an IFRAME (or 507 a redirect) to http://calling-service.example.com 509 3. The attacker forges the response apparently http://calling- 510 service.example.com/ to inject JS to initiate a call to himself. 512 Note that this attack does not depend on the media being insecure. 513 Because the call is to the attacker, it is also encrypted to him. 514 Moreover, it need not be executed immediately; the attacker can 515 "infect" the origin semi-permanently (e.g., with a web worker or a 516 popped-up window that is hidden under the main window.) and thus be 517 able to bug me long after I have left the infected network. This 518 risk is created by allowing calls at all from a page fetched over 519 HTTP. 521 Even if calls are only possible from HTTPS sites, if the site embeds 522 active content (e.g., JavaScript) that is fetched over HTTP or from 523 an untrusted site, because that JavaScript is executed in the 524 security context of the page [finer-grained]. Thus, it is also 525 dangerous to allow WebRTC functionality from HTTPS origins that embed 526 mixed content. Note: this issue is not restricted to PAGES which 527 contain mixed content. If a page from a given origin ever loads 528 mixed content then it is possible for a network attacker to infect 529 the browser's notion of that origin semi-permanently. 531 4.2. Communications Consent Verification 533 As discussed in Section 3.3, allowing web applications unrestricted 534 network access via the browser introduces the risk of using the 535 browser as an attack platform against machines which would not 536 otherwise be accessible to the malicious site, for instance because 537 they are topologically restricted (e.g., behind a firewall or NAT). 538 In order to prevent this form of attack as well as cross-protocol 539 attacks it is important to require that the target of traffic 540 explicitly consent to receiving the traffic in question. Until that 541 consent has been verified for a given endpoint, traffic other than 542 the consent handshake MUST NOT be sent to that endpoint. 544 Note that consent verification is not sufficient to prevent overuse 545 of network resources. Because WebRTC allows for a Web site to create 546 data flows between two browser instances without user consent, it is 547 possible for a malicious site to chew up a signficant amount of a 548 user's bandwidth without incurring significant costs to himself by 549 setting up such a channel to another user. However, as a practical 550 matter there are a large number of Web sites which can act as data 551 sources, so an attacker can at least use downlink bandwidth with 552 existing Web APIs. However, this potential DoS vector reinforces the 553 need for adequate congestion control for WebRTC protocols to ensure 554 that they play fair with other demands on the user's bandwidth. 556 4.2.1. ICE 558 Verifying receiver consent requires some sort of explicit handshake, 559 but conveniently we already need one in order to do NAT hole- 560 punching. ICE [RFC5245] includes a handshake designed to verify that 561 the receiving element wishes to receive traffic from the sender. It 562 is important to remember here that the site initiating ICE is 563 presumed malicious; in order for the handshake to be secure the 564 receiving element MUST demonstrate receipt/knowledge of some value 565 not available to the site (thus preventing the site from forging 566 responses). In order to achieve this objective with ICE, the STUN 567 transaction IDs must be generated by the browser and MUST NOT be made 568 available to the initiating script, even via a diagnostic interface. 570 Verifying receiver consent also requires verifying the receiver wants 571 to receive traffic from a particular sender, and at this time; for 572 example a malicious site may simply attempt ICE to known servers that 573 are using ICE for other sessions. ICE provides this verification as 574 well, by using the STUN credentials as a form of per-session shared 575 secret. Those credentials are known to the Web application, but 576 would need to also be known and used by the STUN-receiving element to 577 be useful. 579 There also needs to be some mechanism for the browser to verify that 580 the target of the traffic continues to wish to receive it. Because 581 ICE keepalives are indications, they will not work here. 582 [I-D.ietf-rtcweb-stun-consent-freshness] describes the mechanism for 583 providing consent freshness. 585 4.2.2. Masking 587 Once consent is verified, there still is some concern about 588 misinterpretation attacks as described by Huang et al.[huang-w2sp]. 589 Where TCP is used the risk is substantial due to the potential 590 presence of transparent proxies and therefore if TCP is to be used, 591 then WebSockets style masking MUST be employed. 593 Since DTLS (with the anti-chosen plaintext mechanisms required by TLS 594 1.1) does not allow the attacker to generate predictable ciphertext, 595 there is no need for masking of protocols running over DTLS (e.g. 596 SCTP over DTLS, UDP over DTLS, etc.). 598 Note that in principle an attacker could exert some control over SRTP 599 packets by using a combination of the WebAudio API and extremely 600 tight timing control. The primary risk here seems to be carriage of 601 SRTP over TURN TCP. However, as SRTP packets have an extremely 602 characteristic packet header it seems unlikely that any but the most 603 aggressive intermediaries would be confused into thinking that 604 another application layer protocol was in use. 606 4.2.3. Backward Compatibility 608 A requirement to use ICE limits compatibility with legacy non-ICE 609 clients. It seems unsafe to completely remove the requirement for 610 some check. All proposed checks have the common feature that the 611 browser sends some message to the candidate traffic recipient and 612 refuses to send other traffic until that message has been replied to. 613 The message/reply pair must be generated in such a way that an 614 attacker who controls the Web application cannot forge them, 615 generally by having the message contain some secret value that must 616 be incorporated (e.g., echoed, hashed into, etc.). Non-ICE 617 candidates for this role (in cases where the legacy endpoint has a 618 public address) include: 620 o STUN checks without using ICE (i.e., the non-RTC-web endpoint sets 621 up a STUN responder.) 623 o Use or RTCP as an implicit reachability check. 625 In the RTCP approach, the WebRTC endpoint is allowed to send a 626 limited number of RTP packets prior to receiving consent. This 627 allows a short window of attack. In addition, some legacy endpoints 628 do not support RTCP, so this is a much more expensive solution for 629 such endpoints, for which it would likely be easier to implement ICE. 630 For these two reasons, an RTCP-based approach does not seem to 631 address the security issue satisfactorily. 633 In the STUN approach, the WebRTC endpoint is able to verify that the 634 recipient is running some kind of STUN endpoint but unless the STUN 635 responder is integrated with the ICE username/password establishment 636 system, the WebRTC endpoint cannot verify that the recipient consents 637 to this particular call. This may be an issue if existing STUN 638 servers are operated at addresses that are not able to handle 639 bandwidth-based attacks. Thus, this approach does not seem 640 satisfactory either. 642 If the systems are tightly integrated (i.e., the STUN endpoint 643 responds with responses authenticated with ICE credentials) then this 644 issue does not exist. However, such a design is very close to an 645 ICE-Lite implementation (indeed, arguably is one). An intermediate 646 approach would be to have a STUN extension that indicated that one 647 was responding to WebRTC checks but not computing integrity checks 648 based on the ICE credentials. This would allow the use of standalone 649 STUN servers without the risk of confusing them with legacy STUN 650 servers. If a non-ICE legacy solution is needed, then this is 651 probably the best choice. 653 Once initial consent is verified, we also need to verify continuing 654 consent, in order to avoid attacks where two people briefly share an 655 IP (e.g., behind a NAT in an Internet cafe) and the attacker arranges 656 for a large, unstoppable, traffic flow to the network and then 657 leaves. The appropriate technologies here are fairly similar to 658 those for initial consent, though are perhaps weaker since the 659 threats is less severe. 661 4.2.4. IP Location Privacy 663 Note that as soon as the callee sends their ICE candidates, the 664 caller learns the callee's IP addresses. The callee's server 665 reflexive address reveals a lot of information about the callee's 666 location. In order to avoid tracking, implementations may wish to 667 suppress the start of ICE negotiation until the callee has answered. 668 In addition, either side may wish to hide their location entirely by 669 forcing all traffic through a TURN server. 671 In ordinary operation, the site learns the browser's IP address, 672 though it may be hidden via mechanisms like Tor 673 [http://www.torproject.org] or a VPN. However, because sites can 674 cause the browser to provide IP addresses, this provides a mechanism 675 for sites to learn about the user's network environment even if the 676 user is behind a VPN that masks their IP address. Implementations 677 may wish to provide settings which suppress all non-VPN candidates if 678 the user is on certain kinds of VPN, especially privacy-oriented 679 systems such as Tor. 681 4.3. Communications Security 683 Finally, we consider a problem familiar from the SIP world: 684 communications security. For obvious reasons, it MUST be possible 685 for the communicating parties to establish a channel which is secure 686 against both message recovery and message modification. (See 687 [RFC5479] for more details.) This service must be provided for both 688 data and voice/video. Ideally the same security mechanisms would be 689 used for both types of content. Technology for providing this 690 service (for instance, SRTP [RFC3711], DTLS [RFC4347] and DTLS-SRTP 691 [RFC5763]) is well understood. However, we must examine this 692 technology to the WebRTC context, where the threat model is somewhat 693 different. 695 In general, it is important to understand that unlike a conventional 696 SIP proxy, the calling service (i.e., the Web server) controls not 697 only the channel between the communicating endpoints but also the 698 application running on the user's browser. While in principle it is 699 possible for the browser to cut the calling service out of the loop 700 and directly present trusted information (and perhaps get consent), 701 practice in modern browsers is to avoid this whenever possible. "In- 702 flow" modal dialogs which require the user to consent to specific 703 actions are particularly disfavored as human factors research 704 indicates that unless they are made extremely invasive, users simply 705 agree to them without actually consciously giving consent. 706 [abarth-rtcweb]. Thus, nearly all the UI will necessarily be 707 rendered by the browser but under control of the calling service. 709 This likely includes the peer's identity information, which, after 710 all, is only meaningful in the context of some calling service. 712 This limitation does not mean that preventing attack by the calling 713 service is completely hopeless. However, we need to distinguish 714 between two classes of attack: 716 Retrospective compromise of calling service. 718 The calling service is is non-malicious during a call but 719 subsequently is compromised and wishes to attack an older call 720 (often called a "passive attack") 722 During-call attack by calling service. 724 The calling service is compromised during the call it wishes to 725 attack (often called an "active attack"). 727 Providing security against the former type of attack is practical 728 using the techniques discussed in Section 4.3.1. However, it is 729 extremely difficult to prevent a trusted but malicious calling 730 service from actively attacking a user's calls, either by mounting a 731 MITM attack or by diverting them entirely. (Note that this attack 732 applies equally to a network attacker if communications to the 733 calling service are not secured.) We discuss some potential 734 approaches and why they are likely to be impractical in 735 Section 4.3.2. 737 4.3.1. Protecting Against Retrospective Compromise 739 In a retrospective attack, the calling service was uncompromised 740 during the call, but that an attacker subsequently wants to recover 741 the content of the call. We assume that the attacker has access to 742 the protected media stream as well as having full control of the 743 calling service. 745 If the calling service has access to the traffic keying material (as 746 in SDES [RFC4568]), then retrospective attack is trivial. This form 747 of attack is particularly serious in the Web context because it is 748 standard practice in Web services to run extensive logging and 749 monitoring. Thus, it is highly likely that if the traffic key is 750 part of any HTTP request it will be logged somewhere and thus subject 751 to subsequent compromise. It is this consideration that makes an 752 automatic, public key-based key exchange mechanism imperative for 753 WebRTC (this is a good idea for any communications security system) 754 and this mechanism SHOULD provide perfect forward secrecy (PFS). The 755 signaling channel/calling service can be used to authenticate this 756 mechanism. 758 In addition, if end-to-end keying is in used, the system MUST NOT 759 provide any APIs to extract either long-term keying material or to 760 directly access any stored traffic keys. Otherwise, an attacker who 761 subsequently compromised the calling service might be able to use 762 those APIs to recover the traffic keys and thus compromise the 763 traffic. 765 4.3.2. Protecting Against During-Call Attack 767 Protecting against attacks during a call is a more difficult 768 proposition. Even if the calling service cannot directly access 769 keying material (as recommended in the previous section), it can 770 simply mount a man-in-the-middle attack on the connection, telling 771 Alice that she is calling Bob and Bob that he is calling Alice, while 772 in fact the calling service is acting as a calling bridge and 773 capturing all the traffic. Protecting against this form of attack 774 requires positive authentication of the remote endpoint such as 775 explicit out-of-band key verification (e.g., by a fingerprint) or a 776 third-party identity service as described in 777 [I-D.ietf-rtcweb-security-arch]. 779 4.3.2.1. Key Continuity 781 One natural approach is to use "key continuity". While a malicious 782 calling service can present any identity it chooses to the user, it 783 cannot produce a private key that maps to a given public key. Thus, 784 it is possible for the browser to note a given user's public key and 785 generate an alarm whenever that user's key changes. SSH [RFC4251] 786 uses a similar technique. (Note that the need to avoid explicit user 787 consent on every call precludes the browser requiring an immediate 788 manual check of the peer's key). 790 Unfortunately, this sort of key continuity mechanism is far less 791 useful in the WebRTC context. First, much of the virtue of WebRTC 792 (and any Web application) is that it is not bound to particular piece 793 of client software. Thus, it will be not only possible but routine 794 for a user to use multiple browsers on different computers which will 795 of course have different keying material (SACRED [RFC3760] 796 notwithstanding.) Thus, users will frequently be alerted to key 797 mismatches which are in fact completely legitimate, with the result 798 that they are trained to simply click through them. As it is known 799 that users routinely will click through far more dire warnings 800 [cranor-wolf], it seems extremely unlikely that any key continuity 801 mechanism will be effective rather than simply annoying. 803 Moreover, it is trivial to bypass even this kind of mechanism. 804 Recall that unlike the case of SSH, the browser never directly gets 805 the peer's identity from the user. Rather, it is provided by the 806 calling service. Even enabling a mechanism of this type would 807 require an API to allow the calling service to tell the browser "this 808 is a call to user X". All the calling service needs to do to avoid 809 triggering a key continuity warning is to tell the browser that "this 810 is a call to user Y" where Y is close to X. Even if the user 811 actually checks the other side's name (which all available evidence 812 indicates is unlikely), this would require (a) the browser to trusted 813 UI to provide the name and (b) the user to not be fooled by similar 814 appearing names. 816 4.3.2.2. Short Authentication Strings 818 ZRTP [RFC6189] uses a "short authentication string" (SAS) which is 819 derived from the key agreement protocol. This SAS is designed to be 820 compared by the users (e.g., read aloud over the the voice channel or 821 transmitted via an out of band channel) and if confirmed by both 822 sides precludes MITM attack. The intention is that the SAS is used 823 once and then key continuity (though a different mechanism from that 824 discussed above) is used thereafter. 826 Unfortunately, the SAS does not offer a practical solution to the 827 problem of a compromised calling service. "Voice conversion" 828 systems, which modify voice from one speaker to make it sound like 829 another, are an active area of research. These systems are already 830 good enough to fool both automatic recognition systems 831 [farus-conversion] and humans [kain-conversion] in many cases, and 832 are of course likely to improve in future, especially in an 833 environment where the user just wants to get on with the phone call. 834 Thus, even if SAS is effective today, it is likely not to be so for 835 much longer. 837 Additionally, it is unclear that users will actually use an SAS. As 838 discussed above, the browser UI constraints preclude requiring the 839 SAS exchange prior to completing the call and so it must be 840 voluntary; at most the browser will provide some UI indicator that 841 the SAS has not yet been checked. However, it it is well-known that 842 when faced with optional security mechanisms, many users simply 843 ignore them [whitten-johnny]. 845 Once users have checked the SAS once, key continuity is required to 846 avoid them needing to check it on every call. However, this is 847 problematic for reasons indicated in Section 4.3.2.1. In principle 848 it is of course possible to render a different UI element to indicate 849 that calls are using an unauthenticated set of keying material 850 (recall that the attacker can just present a slightly different name 851 so that the attack shows the same UI as a call to a new device or to 852 someone you haven't called before) but as a practical matter, users 853 simply ignore such indicators even in the rather more dire case of 854 mixed content warnings. 856 4.3.2.3. Third Party Identity 858 The conventional approach to providing communications identity has of 859 course been to have some third party identity system (e.g., PKI) to 860 authenticate the endpoints. Such mechanisms have proven to be too 861 cumbersome for use by typical users (and nearly too cumbersome for 862 administrators). However, a new generation of Web-based identity 863 providers (BrowserID, Federated Google Login, Facebook Connect, 864 OAuth, OpenID, WebFinger), has recently been developed and use Web 865 technologies to provide lightweight (from the user's perspective) 866 third-party authenticated transactions. It is possible to use 867 systems of this type to authenticate WebRTC calls, linking them to 868 existing user notions of identity (e.g., Facebook adjacencies). 869 Specifically, the third-party identity system is used to bind the 870 user's identity to cryptographic keying material which is then used 871 to authenticate the calling endpoints. Calls which are authenticated 872 in this fashion are naturally resistant even to active MITM attack by 873 the calling site. 875 Note that there is one special case in which PKI-style certificates 876 do provide a practical solution: calls from end-users to large sites. 877 For instance, if you are making a call to Amazon.com, then Amazon can 878 easily get a certificate to authenticate their media traffic, just as 879 they get one to authenticate their Web traffic. This does not 880 provide additional security value in cases in which the calling site 881 and the media peer are one in the same, but might be useful in cases 882 in which third parties (e.g., ad networks or retailers) arrange for 883 calls but do not participate in them. 885 4.3.2.4. Page Access to Media 887 Identifying the identity of the far media endpoint is a necessary but 888 not sufficient condition for providing media security. In WebRTC, 889 media flows are rendered into HTML5 MediaStreams which can be 890 manipulated by the calling site. Obviously, if the site can modify 891 or view the media, then the user is not getting the level of 892 assurance they would expect from being able to authenticate their 893 peer. In many cases, this is acceptable because the user values 894 site-based special effects over complete security from the site. 895 However, there are also cases where users wish to know that the site 896 cannot interfere. In order to facilitate that, it will be necessary 897 to provide features whereby the site can verifiably give up access to 898 the media streams. This verification must be possible both from the 899 local side and the remote side. I.e., I must be able to verify that 900 the person I am calling has engaged a secure media mode. In order to 901 achieve this it will be necessary to cryptographically bind an 902 indication of the local media access policy into the cryptographic 903 authentication procedures detailed in the previous sections. 905 4.3.3. Malicious Peers 907 One class of attack that we do not generally try to prevent is 908 malicious peers. For instance, no matter what confidentiality 909 measures you employ the person you are talking to might record the 910 call and publish it on the Internet. Similarly, we do not attempt to 911 prevent them from using voice or video processing technology from 912 hiding or changing their appearance. While technologies (DRM, etc.) 913 do exist to attempt to address these issues, they are generally not 914 compatible with open systems and WebRTC does not address them. 916 Similarly, we make no attempt to prevent prank calling or other 917 unwanted calls. In general, this is in the scope of the calling 918 site, though because WebRTC does offer some forms of strong 919 authentication, that may be useful as part of a defense against such 920 attacks. 922 4.4. Privacy Considerations 924 4.4.1. Correlation of Anonymous Calls 926 While persistent endpoint identifiers can be a useful security 927 feature (see Section 4.3.2.1 they can also represent a privacy threat 928 in settings where the user wishes to be anonymous. WebRTC provides a 929 number of possible persistent identifiers such as DTLS certificates 930 (if they are reused between connections) and RTCP CNAMES (if 931 generated according to [RFC6222] rather than the privacy preserving 932 mode of [I-D.ietf-avtcore-6222bis]). In order to prevent this type 933 of correlation, browsers need to provide mechanisms to reset these 934 identifiers (e.g., with the same lifetime as cookies). Moreover, the 935 API should provide mechanisms to allow sites intended for anonymous 936 calling to force the minting of fresh identifiers. In addition, IP 937 addresses can be a source of call linkage 938 [I-D.ietf-rtcweb-ip-handling] 940 4.4.2. Browser Fingerprinting 942 Any new set of API features adds a risk of browser fingerprinting, 943 and WebRTC is no exception. Specifically, sites can use the presence 944 or absence of specific devices as a browser fingerprint. In general, 945 the API needs to be balanced between functionality and the 946 incremental fingerprint risk. 948 5. Security Considerations 950 This entire document is about security. 952 6. Acknowledgements 954 Bernard Aboba, Harald Alvestrand, Dan Druta, Cullen Jennings, Alan 955 Johnston, Hadriel Kaplan (S 4.2.1), Matthew Kaufman, Martin Thomson, 956 Magnus Westerlund. 958 7. Changes Since -04 960 o Replaced RTCWEB and RTC-Web with WebRTC, except when referring to 961 the IETF WG 963 o Removed discussion of the IFRAMEd advertisement case, since we 964 decided not to treat it specially. 966 o Added a privacy section considerations section. 968 o Significant edits to the SAS section to reflect Alan Johnston's 969 comments. 971 o Added some discussion if IP location privacy and Tor. 973 o Updated the "communications consent" section to reflrect draft- 974 ietf. 976 o Added a section about "malicious peers". 978 o Added a section describing screen sharing threats. 980 o Assorted editorial changes. 982 8. References 984 8.1. Normative References 986 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 987 Requirement Levels", BCP 14, RFC 2119, 988 DOI 10.17487/RFC2119, March 1997, . 991 8.2. Informative References 993 [abarth-rtcweb] 994 Barth, A., "Prompting the user is security failure", RTC- 995 Web Workshop, September 2010. 997 [CORS] van Kesteren, A., "Cross-Origin Resource Sharing", January 998 2014. 1000 [cranor-wolf] 1001 Sunshine, J., Egelman, S., Almuhimedi, H., Atri, N., and 1002 L. cranor, "Crying Wolf: An Empirical Study of SSL Warning 1003 Effectiveness", Proceedings of the 18th USENIX Security 1004 Symposium, 2009, August 2009. 1006 [farus-conversion] 1007 Farrus, M., Erro, D., and J. Hernando, "Speaker 1008 Recognition Robustness to Voice Conversion", January 2008. 1010 [finer-grained] 1011 Barth, A. and C. Jackson, "Beware of Finer-Grained 1012 Origins", W2SP, 2008, July 2008. 1014 [huang-w2sp] 1015 Huang, L-S., Chen, E., Barth, A., Rescorla, E., and C. 1016 Jackson, "Talking to Yourself for Fun and Profit", W2SP, 1017 2011, May 2011. 1019 [I-D.ietf-avtcore-6222bis] 1020 Begen, A., Perkins, C., Wing, D., and E. Rescorla, 1021 "Guidelines for Choosing RTP Control Protocol (RTCP) 1022 Canonical Names (CNAMEs)", draft-ietf-avtcore-6222bis-06 1023 (work in progress), July 2013. 1025 [I-D.ietf-rtcweb-ip-handling] 1026 Uberti, J. and G. Shieh, "WebRTC IP Address Handling 1027 Requirements", draft-ietf-rtcweb-ip-handling-04 (work in 1028 progress), July 2017. 1030 [I-D.ietf-rtcweb-overview] 1031 Alvestrand, H., "Overview: Real Time Protocols for 1032 Browser-based Applications", draft-ietf-rtcweb-overview-18 1033 (work in progress), March 2017. 1035 [I-D.ietf-rtcweb-security-arch] 1036 Rescorla, E., "WebRTC Security Architecture", draft-ietf- 1037 rtcweb-security-arch-12 (work in progress), June 2016. 1039 [I-D.ietf-rtcweb-stun-consent-freshness] 1040 Perumal, M., Wing, D., R, R., Reddy, T., and M. Thomson, 1041 "STUN Usage for Consent Freshness", draft-ietf-rtcweb- 1042 stun-consent-freshness-16 (work in progress), August 2015. 1044 [I-D.kaufman-rtcweb-security-ui] 1045 Kaufman, M., "Client Security User Interface Requirements 1046 for RTCWEB", draft-kaufman-rtcweb-security-ui-00 (work in 1047 progress), June 2011. 1049 [kain-conversion] 1050 Kain, A. and M. Macon, "Design and Evaluation of a Voice 1051 Conversion Algorithm based on Spectral Envelope Mapping 1052 and Residual Prediction", Proceedings of ICASSP, May 1053 2001, May 2001. 1055 [RFC2818] Rescorla, E., "HTTP Over TLS", RFC 2818, 1056 DOI 10.17487/RFC2818, May 2000, . 1059 [RFC3261] Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston, 1060 A., Peterson, J., Sparks, R., Handley, M., and E. 1061 Schooler, "SIP: Session Initiation Protocol", RFC 3261, 1062 DOI 10.17487/RFC3261, June 2002, . 1065 [RFC3552] Rescorla, E. and B. Korver, "Guidelines for Writing RFC 1066 Text on Security Considerations", BCP 72, RFC 3552, 1067 DOI 10.17487/RFC3552, July 2003, . 1070 [RFC3711] Baugher, M., McGrew, D., Naslund, M., Carrara, E., and K. 1071 Norrman, "The Secure Real-time Transport Protocol (SRTP)", 1072 RFC 3711, DOI 10.17487/RFC3711, March 2004, 1073 . 1075 [RFC3760] Gustafson, D., Just, M., and M. Nystrom, "Securely 1076 Available Credentials (SACRED) - Credential Server 1077 Framework", RFC 3760, DOI 10.17487/RFC3760, April 2004, 1078 . 1080 [RFC4251] Ylonen, T. and C. Lonvick, Ed., "The Secure Shell (SSH) 1081 Protocol Architecture", RFC 4251, DOI 10.17487/RFC4251, 1082 January 2006, . 1084 [RFC4347] Rescorla, E. and N. Modadugu, "Datagram Transport Layer 1085 Security", RFC 4347, DOI 10.17487/RFC4347, April 2006, 1086 . 1088 [RFC4568] Andreasen, F., Baugher, M., and D. Wing, "Session 1089 Description Protocol (SDP) Security Descriptions for Media 1090 Streams", RFC 4568, DOI 10.17487/RFC4568, July 2006, 1091 . 1093 [RFC5245] Rosenberg, J., "Interactive Connectivity Establishment 1094 (ICE): A Protocol for Network Address Translator (NAT) 1095 Traversal for Offer/Answer Protocols", RFC 5245, 1096 DOI 10.17487/RFC5245, April 2010, . 1099 [RFC5479] Wing, D., Ed., Fries, S., Tschofenig, H., and F. Audet, 1100 "Requirements and Analysis of Media Security Management 1101 Protocols", RFC 5479, DOI 10.17487/RFC5479, April 2009, 1102 . 1104 [RFC5763] Fischl, J., Tschofenig, H., and E. Rescorla, "Framework 1105 for Establishing a Secure Real-time Transport Protocol 1106 (SRTP) Security Context Using Datagram Transport Layer 1107 Security (DTLS)", RFC 5763, DOI 10.17487/RFC5763, May 1108 2010, . 1110 [RFC6189] Zimmermann, P., Johnston, A., Ed., and J. Callas, "ZRTP: 1111 Media Path Key Agreement for Unicast Secure RTP", 1112 RFC 6189, DOI 10.17487/RFC6189, April 2011, 1113 . 1115 [RFC6222] Begen, A., Perkins, C., and D. Wing, "Guidelines for 1116 Choosing RTP Control Protocol (RTCP) Canonical Names 1117 (CNAMEs)", RFC 6222, DOI 10.17487/RFC6222, April 2011, 1118 . 1120 [RFC6454] Barth, A., "The Web Origin Concept", RFC 6454, 1121 DOI 10.17487/RFC6454, December 2011, . 1124 [RFC6455] Fette, I. and A. Melnikov, "The WebSocket Protocol", 1125 RFC 6455, DOI 10.17487/RFC6455, December 2011, 1126 . 1128 [SWF] Adobe, "SWF File Format Specification Version 19", April 1129 2013. 1131 [whitten-johnny] 1132 Whitten, A. and J. Tygar, "Why Johnny Can't Encrypt: A 1133 Usability Evaluation of PGP 5.0", Proceedings of the 8th 1134 USENIX Security Symposium, 1999, August 1999. 1136 Author's Address 1137 Eric Rescorla 1138 RTFM, Inc. 1139 2064 Edgewood Drive 1140 Palo Alto, CA 94303 1141 USA 1143 Phone: +1 650 678 2350 1144 Email: ekr@rtfm.com