Content Cruising System under Dense Region of Mobile Nodes - PowerPoint PPT Presentation

slide1 n.
Download
Skip this Video
Loading SlideShow in 5 Seconds..
Content Cruising System under Dense Region of Mobile Nodes PowerPoint Presentation
Download Presentation
Content Cruising System under Dense Region of Mobile Nodes

play fullscreen
1 / 82
Content Cruising System under Dense Region of Mobile Nodes
143 Views
Download Presentation
miya
Download Presentation

Content Cruising System under Dense Region of Mobile Nodes

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

  1. Content Cruising System under Dense Region of Mobile Nodes Takaaki Ishida*, Masayoshi Imaike** * Research Institute for Digital Media and Content, Keio University ** FEAC International Corp. *ishi@dmc.keio.ac.jp, **isle@feac.jp

  2. Introduction

  3. About Location-based Service The service distributing contents to users depending on their positional information is called location based service. “ROAD CONSTRUCTION, NEXT 4 MILES” “You can find a McDonald's around here!” Source of image: KDDI Corporation JAPAN

  4. Topics of our Research • Research Motivation: • To realize free and robust location based services that do not necessarily depend on specific infrastructures. • Keywords: • Accidental encountering the valuable information through Wireless P2P ad-hoc communication • Adopting simple “Store & Forward” transmission model instead of complicated routing protocol • Generation of new communication model by utilizing of “Physical Movement “of mobile nodes • We are developing “Content Cruising System“ which is decentralized in form that mobile nodes distributed in various places sharing the functions to store and to distribute the contents.

  5. Content Cruising System

  6. Overview Receivers & Mediators prompt the content to be transferred to the destination point, .and to be remained around the destination area during the duration by cooperation of scattered mobile nodes. Sender: 1) adds the metadata which describes “Destination point”, “Duration”, and “Content ID” to the content. 2) then send it to one-hop neighbors which happen to meet. Content Cruising System metadata Content

  7. Operation of CCS Servent PDA a PDA b Presentation GPS GPS Presentation Presentation Algorithm Presentation Algorithm SelectionAlgorithm Receive Send TransmissionAlgorithm Frequency ControlAlgorithm Timer Algorithm ContextManager ContextManager Storage IPv6Link-local all node multicast IPv6Link-local all node multicast Storage Frequency ControlAlgorithm TransmissionAlgorithm SelectionAlgorithm Timer Algorithm Send Receive

  8. Configuration of Transmission Algorithm “d” is the distance from the destination point, “a” is a coefficient, . ”k” is the parameter which define the dependence with distance. Centripetal Force Interval Time of Sending a 2a 1 -kd a y = ae y -1 =a -1 e kd 2 a R R (0.0) (0.0)

  9. Two autonomous control mechanisms under Dense Region of Mobile Nodes

  10. Concerns about Scalability • Two unforeseen phenomena which cause a malfunction under dense region of mobile nodes • Increase in the number of mobile nodes • Increase in the number of of contents • To avoid these problems, CCS has autonomous control mechanisms which are constructed by cooperation of several modules and algorithms.

  11. Increase in the number of nodes Mobile nodes connected to same link transmission interval = 3sec transmission interval = 3sec transmission interval = 2sec Destination point data data data data data data data data data data data data data data data data data Total amount of network traffics on same link

  12. Function of Timer reset Destination of the content Only the node which is nearest to the destination point in same link sends the content

  13. Autonomous Control Mechanism to the Increase of mobile nodes Mobile nodes connected to same link transmission interval = 3sec transmission interval = 3sec transmission interval = 2sec data data data reset reset reset reset reset reset data data data data data data data reset reset reset reset reset reset reset data data data reset data data data data data data data Only the node which is nearest to the destination point sends the content

  14. Increase in the number of contents Total amount of network traffics on same link are: node 1 node 2 Traffics consumed by node1 Traffics consumed by node2 ↑transmitting ratio of each content distance from the destination → ←distance from the destination

  15. Function of Frequency Control Algorithm The following information is added to the header of each content exchanged by CCS: -data size (byte) -default transmission frequency (sec) -congestion control coefficient (decimal) T = data size / default transmission frequency Network traffic on same link Tmax Each node calculates “C”. according to header info and reconfigures the transmission frequency of each content multiplying it by this value of “C”. Increasing of number of content

  16. Verification Experiments

  17. experimental environment Within same radio range Environment setting • 5 sets of PDAs which CCS servant application was installed were placed in effective radius of wireless LAN. • One node was located at the center point of all, and the other four nodes had the implicit latitude and longitude set up as if they were placed in a line every other meter from the center node. • Whenever the node was one meter away from the destination point of content, Transmitting Frequency is set up so that it might shift by a unit of one second. 1m Allocation points of PDAs (Virtual latitude and longitude) Destination point of content

  18. Verification experiment of Timer Algorithm • A dummy content (data size: 1Kbyte, destination: same position with the center node, default transmission frequency at the destination: once per 5 sec) was sent to the five nodes, and the number of transmissions in each nodes were counted for duration of 6 minutes. • The experiment was conducted in two cases whether the timer algorithm is applied or not.

  19. Experimental Result of Timer Algorithm

  20. Verification experiment of Frequency Control Algorithm • The congestion control coefficient was set up as the restriction of bandwidth to be 2.5Kbyte/sec. • Dummy contents (data size: 1Kbyte, destination: same position of the center node, default transmission frequency at the destination: once per 2 sec) were sent to the five nodes. • The experiments were conducted five times with numbers of content being changed (the restriction of bandwidth was multiplied by the experiment number each time), and total traffic which flows on the link was measured in 5 minutes per experiment. • These experiments were conducted in two cases (with or without Frequency Control Algorithm) in same condition, and both average traffics were compared.

  21. Experimental Result of Frequency Control Algorithm

  22. Summary • To make the CSS work properly under the severe environment dense with mobile nodes, it is a critical issue to find a solution to handle the increasing nodes and contents. The "Timer algorithm" and "Frequency control algorithm" are developed as autonomous control mechanisms in order to avoid malfunction in such a situation. • Both of these mechanisms worked well through the experiments by using of several PDAs. Hereafter, we will continue to verify practical effectiveness of these mechanisms assuming more realistic situation by using of a simulator.

  23. Thank you for listening

  24. Current Operation of Location-based Service • It requires: • Costly equipments (numerous base stations, Content storage/distribution servers, etc.) • Central management. • Continuous network connectivity . • The problem of single point of failure arises • We cannot use it: • in some developing region without such an infrastructure • in a disastrous situation where the infrastructure is damaged Internet GPS Base station Data Base Send Location information as a query areaA Get it’s Location information Retrieve content of it’s area from DB area A

  25. 1m タイマーアルゴリズムの検証 1hopの距離 • 実験内容 • 10台の愛・MATEを無線LANの有効 半径内に設置 • 中心ノードの緯度経度を固定 • 中心ノードから実測1m置きに並んでいるように、他の9台の端末の緯度経度を固定※右図参照 • 送信頻度を以下のように定め、中心ノードから約5KBのデータを発信してタイマーアルゴリズムが機能しているかを調べた。 • T=α+X^β×φ ※今回の実験では、パラメータを以下のように定め、送信頻度Tは1秒以下を切捨てとした。 α=3sec,X=1m,β=2,Φ=0.2 • 上記の計算式により、各設置端末の送信インターバルは理論上下記の数値となる。 • 比較検証 • 同じ条件で、タイマーアルゴリズムを入れた場合と入れない場合の送信回数を比較した。 端末の設置位置(仮想の緯度経度) コンテンツの目的地

  26. 隠れ端末を想定した場合 ×2 1 2 3 4 5 1 2 4 5 3 1 2 3 5 4 共有リンク上 での送出間隔 1 1 2 2 3 3 4 4 5 6 1 2 3 4 1 2 3 5 4

  27. About each Algorithm • Transmission Algorithm: • Objectives:To change the frequency of transmission depending on the relation between their location and content’s destination. • Selection Algorithm: • Objectives:To avoid the depuration of content which it received. • Operation:It compares the Content ID of the new content with those of the old ones in order to detect the duplication.If duplication is detected, this module discards the old content . • Timer Algorithm: • Objectives:To set the timer to next transmission on each content depending on Transmission Algorithm. • Operation: • Set the timer to next transmission on each content • Reset a timer of the content when it received same content from others. • Stop resetting a timer randomly. • Presentation Algorithm: • Objectives:To pick up the content whose destination is close to user’s location from Storage Module. • Operation:Display the content whose destination is within the range of 5 meters.

  28. ノード数に対するスケーラビリティーの検証

  29. タイマーリセットを入れた場合 同一リンク上のノードが配信するコンテンツ 3秒ごと 3秒ごと 2秒ごと data data data reset reset reset reset reset reset data data data data data data data reset reset reset reset reset reset reset data data data reset data data data data data data data

  30. ノード数増加に対するスケーラビリティー Destination of the content

  31. ノード数増加に対するスケーラビリティー Destination of the content

  32. Mobile Ad-hoc Communication In the mobile wireless Communication: Mobile nodes have their radio device’s range.(ex. Bluetooth → 10m, IEEE802.11b →100m, etc.) In their radio range, they can: • detect the nodes which are close geographically • broadcast contents to their One-hop neighbors which they happen to meet. • The nodes which receive the content forward it to other nodes. Wireless P2PAd-hoc communication

  33. Merits and Demerits Using the catenation of ad-hoc communication as a location based services generates both merits and demerits. Merits: • Contents are transferred to further areas by physical movements of numerous mobile nodes. • Costly equipments and network infrastructure are not needed. Demerits: • Contents are spread disorderly in wide area through multi-hop communication.

  34. Our Approach Location-based service will be realized If information flooding can be suppressed to fixed geographical areas. To attain the goal which mentioned above: • Each content shall have it’s “Geographical destination” and “Duration” previously. • Content sender adds these information to each content as meta-data. • Mobile nodes which receive the content change frequency of transmitting the content depending on their location information which is obtained by GPS. →Contents are gathered around destination area autonomously by cooperation of mobile nodes.

  35. Content Cruising System

  36. Objectives • Location based transmission: Contents are transferred and sustained in a specified area autonomously. • Location-awareness: Contents are found by users in a relevant location automatically. • Location based selection: Unnecessary contents are eliminated based on the user's location.

  37. 想定する情報伝播モデル • 情報の憑依と地縛 • 発信されたコンテンツは「どこに配布したいか」「どのくらいの時間停留させたいか」といった発信者の配布ポリシーを持ち、近隣のノードに無作為に伝達される。 • コンテンツはノードの物理的な移動に従い各地へと運搬され、実空間上を移動する。また、運搬の過程において、発信されたコンテンツは自身の配布ポリシーにそぐわないノードから自然に離脱する。→ 情報の憑依 • 目的とされる地域に近づいたコンテンツはその地に留まり、地域に新しく入ってきた移動ノードに次々と複製と憑依を繰り返すことによって、その地域に停留していく。→情報の地縛 • このようにコンテンツの配布ポリシーを元に多数のモバイルノードが自律・分散的に協調作業を行なってコンテンツを運搬していくことで、あたかもコンテンツ自体に意思があるかのように送信や消滅等自らの次の行動を決定していくという情報伝達のイメージを想定し、システムの設計を行った。

  38. 設計コンセプト 設計コンセプト • 自律・分散・協調モデル • 統治者が存在しない。あるのは共通のルールのみ。 • シンプルなネットワーク設計 • 相互のネゴシエーションやルーティングを前提としない、一方向のブロードキャスト型モデル。 • Context Aware & Action • ノード同士が互いの状況を感知せず、共通のルールに基づいて自律的にアクションを起こすことによって全体の分散・協調システムとして機能する。 • Sensitive Network • 移動先がどのようなネットワークかをネットワーク自体が感知して、自律的に制御する。 →このような条件で、もし情報の流れをコントロールすることが出来れば実空間上におけるRobustな情報配信システムが実現できる。 共通条件である「位置」と「時間」を基準ルールとして設定。 まずはロケーションベースサービスへの適用を想定した実装を行なっている。

  39. Overview Receivers & Mediators prompt the content to be transferred to the destination point, .and to be remained around the destination area during the duration by cooperation of scattered mobile nodes. Sender: 1) adds the metadata which describes “Destination point”, “Duration”, and “Content ID” to the content. 2) then send it to one-hop neighbors which happen to meet. Content Cruising System Metadata (COMPASS) Content

  40. Architecture of CCS Servent • 通信部分 • IPv6リンクローカルオールノードマルチキャスト • モジュール群 • Context Manager • センサーから環境情報(context)を取得。 • ContextとCOMPASSを照合し、コンテンツのスケジュールを管理する。 • Sendモジュール • One Hop Neighborsにコンテンツを同報する • Receiveモジュール • One Hop Neighborsからコンテンツを受信する • Storageモジュール • 受信したコンテンツをコンテンツIDと紐付けて蓄積する • アルゴリズム • Selection Algorithm • 受信したコンテンツの重複チェック(コンテンツIDによる識別) • FIFOオーダーにより、最新情報の生存率を上げる。 • Transmission Algorithm • 送信頻度を計算 • Timer Algorithm • 送信スケジュールの管理 • タイマーリセット機能 • Frequency Control Algorithm • 帯域状況を監視して送信頻度を調整

  41. 万博でのIT実証実験

  42. 全ての交通機関を媒介にした新しいコミュニケーションモデルの実現にむけて全ての交通機関を媒介にした新しいコミュニケーションモデルの実現にむけて

  43. アドホックネットワーク向けルーティング制御プロトコルの分類アドホックネットワーク向けルーティング制御プロトコルの分類 階層型 位置情報補助型 フラット型 CGSRLANMAR ZRP HSR LARGPSR DREAMGeoCast Reactive (On-Demand)型 Proactive (Table-Driven)型 OLSRFSR TBRPFDSDV LANMAR IARP GSR CGSR DSRTORA AODVABR IERPDLAR 複数経路型 ハイブリッド型 ZRP BRP CBRP 太字: RFC化されたプロトコル

  44. ルーティング制御プロトコルの型の分類 

  45. MANET WGで検討中の主なフラット型の     ルーティング制御プロトコル ・DSR (Dynamic Source Routing) ・AODV (Ad hoc On-demand Distance Vector algorithm) ・IERP (IntErzone Routing Protocol) ・DLAR (Dynamic Load-Aware Routing) Reactive型 (On-Demand) ・OLSR (Optimized Link State Routing protocol) ・TBRPF (Topology Broadcast based on Reverse Path Forwarding routing protocol) ・FSR (Fisheye State Routing protocol) ・LANMAR (LANd MARk routing protocol) ・IARP (IntrAzone Routing Protocol) ・DSDV (Destination Sequenced Distance Vector routing) ・GSR (Global State Routing) ・CGSR (Clustered Gateway Switch Routing) Proactive型 (Table-Driven) ・ZRP (Zone Routing Protocol) ・BRP (Bordercast Resolution Protocol) ・CBRP (Cluster-based Routing Protocol) Hybrid型

  46. Reactive型とProactive型の比較 ・インターネットのルーティングプロトコル(RIP, OSPF)はProactive型 ・MACレイヤなのでルーティングとはいえないが、CSMA/CAはReactive型 ・Proactive型では、トポロジーの変更頻度に応じて更新間隔を調整する必要がある    -更新間隔が長すぎると経路情報が古くなり、短すぎるとトラフィックのオーバヘッドが      大きくなる -いかにトポロジー更新情報を効率よく(少ないオーバヘッドで)伝達するかが重要 ・Proactive型では、トポロジーからDijkstraアルゴリズムにより最短経路を決定 

  47. LBSとしての評価(IMSA2003のsimの結果) Graph2 Graph1 Origin (0, 0) is the destination point.X-axis shows the distance from the destination point (graphs show in the range of 800m from the destination point). Y-axis shows the rate of content-reception. The numbers used for the rate of content-reception is calculated by taking the average from 3 different destination points. Graph3

  48. B C A C A B C B A C A 1.Solved by nodes’ movement Star-mark : content’s destination point Node A: First content’s holder When NodeA approaches NodeB and sends the content to NodeB, Node C will be “Glued Node”. • When the glued node’s problem occurs: • The situations the glued node will be able to receive the content are shown below. B B 3: B approaches destination point closer than A 1: B is separated from A 2: A communicates with C directly.