1. 1 Mobile Software Development Framework: Mobile-Cloud Service 10/18/2012 Y. Richard Yang

2. 2 Outline r Admin r Mobile cloud service m Push notification service m Storage service m Track service m Split service

3. 3 Admin. r HW3 posted

4. Recap: Accessing Data in Cloud r A typical design pattern is that a device updates/receives data in the cloud m Cloud as a rendezvous point r Challenge: How do you keep data on a device fresh? 4

5. Recap: Solution Space r Mobile poll r Cloud push r Each app push r Shared (infrastructure) push 5

6. Shared Push Service r A single persistent connection from device to a cloud push service provider r Multiple application providers push to the service provider r Service provider pushes to a device using the persistent connection r Two examples m Apple Push Notification Service (APNS) m Google Cloud Messaging (GCM) 6

7. Design Requirements of a Shared Push Service r Security/Authorization m Do not allow arbitrary app to push to a device r Scalability m A large scale system may have millions of clients r Fault tolerance m Client/device, push servers all can fail r Generality m Can be used by diverse applications 7

8. Design Point: Authorization 8 AppApp Device Registration(DEV_ID, App_ID) Design 2: App query registered devices; Multicast Design 3: Device notifies registration ID to its server; Design 1: App does not know registered devices. Broadcast to all.

9. Design Point: What to Push? r Option 1: Just push signal (data available) to devices and then devices fetch from app servers r Option 2: push app data 9 AppApp Device

10. Design Point: Reliability (What Can Go Wrong) 10 AppApp Device RegID=Registration(DEV_ID, App_ID) App sends to regIDs Device notifies regID to its server;

11. Soft State Design r State at the third party is soft state if the the entity who sets up the state does not refresh it, the state will be pruned at the 3 rd party 11

12. Apple Push Notification Service 12 r iOS device maintains a persistent TCP connection to an Apple Push Notification Server(APNS) A push notification from a provider to a client application Multi-providers to multiple devices

13. APNS Authorization: Device Token 13 r Device token Contains information that enables APNs to locate the device m Client app needs to provide the token to its app provider m Device token should be requested and passed to providers every time your application launches

14. Apple Push Notification Data r Each push notification carries a payload m 256 bytes maximum m Best effort delivery r App provider provides a JSON dictionary object, which contains another dictionary identified by the key aps r aps specifies the following actions An alert message to display to the user A number to badge the application icon with A sound to play 14

15. APNS Example: Client 15 1. - (BOOL)application:(UIApplication *)application didFinishLaunchingWithOptions:(NSDictionary *)launchOptions 2. { 3. // Let the device know we want to receive push notifications 4. [[UIApplication sharedApplication] registerForRemoteNotificationTypes: 5. (UIRemoteNotificationTypeBadge | UIRemoteNotificationTypeSound | UIRemoteNotificationTypeAlert)]; 6. 7. return YES; 8. } 9. - (void)application:(UIApplication*)application didReceiveRemoteNotification:(NSDictionary*)userInfo 10. {//userInfo contains the notification 11. NSLog(@"Received notification: %@", userInfo); 12. } 13. - (void)application:(UIApplication*)application didRegisterForRemoteNotificationsWithDeviceToken:(NSData*)deviceToken 14. { 15. NSLog(@"My token is: %@", deviceToken); 16. }

16. APNS Example: Server 16 1. $devicetoken ='f05571e4be60a4e11524d76e4366862128f430522fb470c46fc6810fffb07af7’; 2. // Put your private key's passphrase here: 3. $passphrase = 'PushChat'; 4. // Put your alert message here: 5. $message = ’CS434: my first push notification!'; 1. $ctx = stream_context_create(); 2. Stream_context_set_option($ctx, 'ssl', 'local_cert', 'ck.pem'); 3. stream_context_set_option($ctx, 'ssl', 'passphrase', $passphrase); 4. // Open a connection to the APNS server 5. $fp = stream_socket_client( 6. 'ssl://gateway.sandbox.push.apple.com:2195', $err, 7. $errstr, 60, STREAM_CLIENT_CONNECT|STREAM_CLIENT_PERSISTENT, $ctx); 8. if (!$fp) 9. exit("Failed to connect: $err $errstr". PHP_EOL); 10. echo 'Connected to APNS'. PHP_EOL;

17. APNS Example: Server (cont’) 17 16. // Create the payload body 17. $body['aps'] = array( 18. 'alert' => $message, 19. 'sound' => 'default' 20. ); 21. // Encode the payload as JSON 22. $payload = json_encode($body); 23. // Build the binary notification 24. $msg = chr(0). pack('n', 32). pack('H*', $deviceToken). pack('n', strlen($payload)). $payload; 25. // Send it to the server 26. $result = fwrite($fp, $msg, strlen($msg)); 27. if (!$result) 28. echo 'Message not delivered'. PHP_EOL; 29. else 30. echo 'Message successfully delivered'. PHP_EOL; 31. // Close the connection to the server 32. fclose($fp);

18. Google Cloud Messaging r Very similar to APNS 18 GCM Servers See http://developer.android.com/guide/google/gcm/gs.html for detailed steps

19. GCM Flow: App Developer Registration r App developer registers a project at Google r Open API console: https://code.google.com/apis/console/https://code.google.com/apis/console/ r After Create project 19 Project ID; Sender ID

20. GCM Flow: Device App Registration r Enable cloud to device messaging in your app m Add permissions in Manifest m App (on device) registers with Google to get registration ID 20 public class MyActivity extends Activity { public void onCreate(Bundle savedInstanceState) { super.onCreate(savedInstanceState); setContentView(R.layout.activity_main); GCMRegistrar.checkDevice(this); GCMRegistrar.checkManifest(this); final String regId = GCMRegistrar.getRegistrationId(this); if (regId.equals("")) { GCMRegistrar.register(this, SENDER_ID); } else { Log.v(TAG, "Already registered"); } … }

21. Device App Handle Events 21  The GCMBroadcastReceiver (defined in GCM library) handles the broadcast messages, and calls methods defined in.GCMIntentService, if you define this service

22. GCMIntentService 22 // called after GCM library finishes registration // you need to send regId to your server onRegistered(Context context, String regId); onUnregistered(Context context, String regId); // called after your server sends a msg to GCM, and // GCM sends to this device onMessage(Context context, Intent intent); onError(Context context, String errorId); onRecoverableError(Context context, String errorId)

23. App Server r If you use GCM server library 23 import com.google.android.gcm.server.*; Sender sender = new Sender(myApiKey); Message message = new Message.Builder().build(); MulticastResult result = sender.send(message, devices, 5);

24. Summary: GCM Flow r Enabling cloud to device messaging m App (on device) registers with Google to get registration ID m App sends registration ID to its App Server r Per message m App Server sends (authenticated) message to Google m Google sends message to device, which sends to app r Disabling cloud to device messaging m App can unregister ID, e.g., when user no longer wants push 24

25. Additional Details r See Thialfi: A Client Notification Service for Internet-Scale Applications, by Atul Adya, Gregory Cooper, Daniel Myers, Michael Piatek, ACM SOSP 2011. 25

26. Discussion: Mobile Cloud Services r We have discussed push notification service. What other services can you think of? 26

27. Example Mobile Cloud Services r Push notification service r Storage and sync m Syncing and storage service (iCloud) r Location based service m Track service (supporting location based services) r Proxy service (Kindle Split Browser) r Recognition services m Speech to text/text to speech service m Natural language processing service (open Siri API for 3 rd party applications in the future) 27

28. Apple iCloud r Store content in cloud and sync to all registered devices m Hosted by Windows Azure and m Amazon AWS r iCloud Storage APIs support third-party app document syncing 28

29. 29 Outline r Admin r Mobile cloud service m Push notification service m Track service StarTrack Next Generation: A Scalable Infrastructure for Track-Based Applications, by Maya Haridasan, Iqbal Mohomed, Doug Terry, Chandramohan A. Thekkath, and Li Zhang, in OSDI 2010.

30. Location-Based Applications r Many phones already have the ability to determine their own location m GPS, cell tower triangulation, or proximity to WiFi hotspots r Many mobile applications use location information Courtesy: Maya et al.

31. A Common Abstraction: Track Time-ordered sequence of location readings Latitude: 37.4013 Longitude: -122.0730 Time: 07/08/10 08:46:45.125 Latitude: 37.4013 Longitude: -122.0730 Time: 07/08/10 08:46:45.125

32. Application: Personalized Driving Directions Goal: Find directions to new gym 32 Courtesy: Maya et al.

33. A Taxonomy of Applications Class of applications enabled by StarTrack 33

34. StarTrack System ST Client Insertion Application Location Manager Retrieval Manipulation Comparison … Application ST Client Insertion ST Server 34

35. System Challenges 1. Handling error-prone tracks 2. Flexible programming interface 3. Efficient implementation of operations on tracks 4. Scalability and fault tolerance 35

36. Challenges of Using Raw Tracks Advantages of Canonicalization: m More efficient retrieval and comparison operations m Enables StarTrack to maintain a list of non-duplicate tracks 36

37. StarTrack API: Track Collections  TC JoinTrackCollections (TC tCs[], bool removeDuplicates)  TC SortTracks (TC tC, SortAttribute attr)  TC TakeTracks(TC tC, int count)  TC GetSimilarTracks (TC tC, Track refTrack, float simThreshold)  TC GetPassByTracks (TC tC, Area[] areas)  TC GetCommonSegments(TC tC, float freqThreshold) Manipulation Creation  TC MakeCollection(GroupCriteria criteria, bool removeDuplicates) 37

38. API Usage: Ride-Sharing Application // get user’s most popular track in the morning TC myTC = MakeCollection(“name = Maya”, [0800 1000], true); TC myPopTC = SortTracks(myTC, FREQ); Track track = GetTracks(myPopTC, 0, 1); // find tracks of all fellow employees TC msTC = MakeCollection(“name.Employer = MS”, [0800 1000], true); // pick tracks from the community most similar to user’s popular track TC similarTC = GetSimilarTracks(msTC, track, 0.8); Track[] similarTracks = GetTracks(similarTC, 0, 20); // Verify if each track is frequently traveled by its respective owner User[] result = FindOwnersOfFrequentTracks(similarTracks);

39. Track Similarity S1-4 S5 S6-7 Tracks A, B s1 s2 s3 s4 s5 Track D s8 s9 Track C s6 s7

40. Summary r The Track abstraction is simple but quite interesting r Think about abstractions in your project 40

41. Amazon Silk Split-Browser r Accelerates web access m Learns user behavior and precache m Intelligently partion work between local and Amazon cloud 41 http://www.extremetech.com/mobile/97587-amazon-silk-bridging-the-gap-between-desktop-and-tablet-web-browsers

42. Example Benefit of Cloud (RTT -> Server) 42

43. Generalization: mCloud r Basic questions m What architecture best supports mcloud? What resources at where m What programming model best support mcloud? 43

44. Example: MAUI 44 Maui server Smartphone Application Client Proxy Profiler Solver Maui Runtime Server Proxy Profiler Solver Maui Runtime Application RPC Maui Controller

45. How Does a Programmer Use MAUI? r Goal: make it dead-simple to MAUI-ify apps m Build app as a standalone phone app m Add.NET attributes to indicate “remoteable” m Follow a simple set of rules

46. Maui server Smartphone Application Client Proxy Profiler Solver Maui Runtime Server Proxy Profiler Solver Maui Runtime Application RPC Maui Controller Intercepts Application Calls Synchronizes State Chooses local or remote Handles Errors Provides runtime information

47. Profiler Callgraph Execution Time State size Network Latency Network Bandwidth Device Profile CPU Cycles Network Power Cost Network Delay Computational Delay Computational Power Cost Computational Delay Annotated Callgraph

48. B 900 mJ 15ms C 5000 mJ 3000 ms 1000mJ 25000 mJ D 15000 mJ 12000 ms 10000 mJ A Computation energy and delay for execution Energy and delay for state transfer A sample callgraph

49. FindMatch 900 mJ InitializeFace Recognizer 5000 mJ 1000mJ 25000 mJ DetectAndExtract Faces 15000 mJ 10000 mJ User Interface Cheaper to do local

50. FindMatch 900 mJ InitializeFace Recognizer 5000 mJ 1000mJ 25000 mJ DetectAndExtract Faces 15000 mJ 10000 mJ User Interface Cheaper to do local

51. FindMatch InitializeFace Recognizer 1000mJ DetectAndExtract Faces User Interface 20900mJ Cheaper to offload

52. MAUI Implementation r Platform m Windows Mobile 6.5 m.NET Framework 3.5 m HTC Fuze Smartphone m Monsoon power monitor r Applications m Chess m Face Recognition m Arcade Game m Voice-based translator

53. Questions r How much can MAUI reduce energy consumption? r How much can MAUI improve performance? r Can MAUI Run Resource-Intensive Applications?

54. Big savings even on 3G An order of magnitude improvement on Wi-Fi Face Recognizer

55. Improvement of around an order of magnitude Face Recognizer

56. Up to 40% energy savings on Wi-Fi Solver would decide not to offload Arcade Game

57. CPU Intensive even on a Core 2 Duo PC Can be run on the phone with MAUI Translator

58.  Adapt to:  Network Bandwidth/Latency Changes  Variability on method’s computational requirements  Experiment:  Modified off the shelf arcade game application  Physics Modeling (homing missiles)  Evaluated under different latency settings

59. DoLevel HandleMissiles DoFrame HandleEnemies HandleBonuses 11KB + missiles missiles *Missiles take around 60 bytes each 11KB + missiles Required state is smaller Complexity increases with # of missiles

60. DoLevel HandleMissiles DoFrame HandleEnemies HandleBonuses *Missiles take around 60 bytes each  Zero Missiles  Low latency (RTT < 10ms) Computation cost is close to zero Offload starting at DoLevel

61. DoLevel HandleMissiles DoFrame HandleEnemies HandleBonuses *Missiles take around 60 bytes each  5 Missiles  Some latency (RTT = 50ms) Most of the computation cost Very expensive to offload everything Little state to offload Only offload Handle Missiles

62. Roadmap r Motivation r MAUI system design m MAUI proxy m MAUI profiler m MAUI solver r Evaluation r Conclusion