A ¡New ¡Approach ¡to ¡Name-‐‒ Based ¡Link-‐‒State ¡Routing ¡ for ¡Information-‐‒Centric ¡ Networks Ehsan Hemmati 1 J.J. Garcia-Luna-Aceves 1,2 1 UC Santa Cruz 2 Palo Alto Research Center ehsan@ce.ucsc.edu
Outline • State of the Art • Related Works • LSCR o Concept o Communication o Operation • Complexity Comparison • Performance
State ¡of ¡the ¡Art ¡in ¡Shortest-‐‒ Path ¡Routing ¡to ¡Content • Problem: Compute the path of minimum cost from each router to each Prefix in the network. • Routing in ICNs is inherently more difficult than routing in the traditional IP networks. • Content objects are cached opportunistically in the network. • Challenges : o Multi-homed instance o Find loop-free paths o Multi-path routing
Related ¡Works • ICN architectures implement one or some of the following mechanisms to constructing a path for acquiring data: o Flooding requests throughout the whole network. o Flooding topology information and the location of publishers. o Using source routes to content. o Creating spanning tree and use publish-subscribe signaling. • Directed Diffusion o Interests are flooded throughout a sensor network • NLSR o Flood the network with topology information as well as instance information.
Related ¡Works NBRP • o Name-prefix reachability is advertised among content routers o Path information is used to avoid permanent loops • CBCB o Establishes a spanning tree of the network o Sends publish-subscribe requests for content • DCR o Routers choose what information to share with their peers o Uses only distance information to calculate shortest path to the nearest copies of the content
Link-‐‒State ¡Content ¡Routing: ¡ Concept ¡ • Every piece of content in the network is a named- data object (NDO) • A set of one or multiple NDOs can be represented by Prefix • A router that has local access to the content is called an Anchor of the prefix • LSCR relies on two basic mechanisms: o Name resolution o Topology-based routing • LSCR creates a lexicographic ordering among neighbors and calculates loop-free routes to the nearest anchor(s) of prefixes.
LSCR ¡-‐‒ Notations l ni : cost of the link from router i to its neighbor n • N i : set of neighbor routers of node i • | i |: lexicographic value of the identifier of router i • k ij : the king anchor for prefix j • S ij : set of valid next hops toward prefix j • rd ip : distance from router i to router p • rd ipn : ¡ distance from router i to router p through • neighbor n RS ip : set of valid next hops toward router p •
LSCR ¡Tables • Link Cost Table ( LT ) o storing the cost of the link from router i to each of its adjacent routers • Forwarding Table ( FT ) o stores the set of valid next hops to reach each router in the network • Prefix Table ( PT ) o stores information about prefixes and their corresponding anchor(s) Routing Table ( RT ) • o stores routing information for each known prefix
Link ¡state ¡advertisements • Link state advertisements (LSAs) are used to create: o The network topology o Mapping schema from prefixes to router names • Router LSA (RLSA) o Advertises topology information. o Flood to whole network. • Anchor LSA (ALSA) o Advertises the existence of name prefixes. o One prefix update per ALSA o Propagated selectively. o “ vFlag ” indicates if the prefix is attached to anchor or detached • Each LSA has a sequence number that is set by the originator of that LSA
LSCR • Routing to nearest instances of destination: o Calculate valid next hops for each router • create the network topology • Run Dijkstra’s SPF algorithm and calculate cost of the path to every destination • Check NOC condition o Select best neighbors from the previous step as valid next hops to the prefix • Maps prefixes to anchors • Determine the king anchor • Check SOC condition
LSCR • Next-Hop Ordering Condition (NOC) to select its neighbor n as valid next hop toward router p if neighbor n: o reports up-to-date information o can reach the destination ( rd ipn < ¡∞ ) • is closer to destination ( rd ipn < ¡rd ip ) or • is at the same distance and has a smaller name ( rd ipn = ¡d ip ∧ |n| ¡< ¡|i| ) • NOC prevents permanent routing loops from being created
LSCR 3 3 2 4 n d v g 2 5 1 4 a o 3 e x q 2 h 2 r b 5 0 4 z s 1 3 p i t 1 3 c y 2 f u 4 2 Sample ¡Network
LSCR 3 3 2 4 n d v g 2 5 1 4 a o 3 e x q 2 h 2 r b 5 0 4 z s Valid Next hops to destination q 1 3 p i t 1 3 c 2 y 2 1 2 3 f u n d v 4 g 2 3 3 Valid Next hops to destination p 2 1 a o 0 e x q 1 h 3 r b 3 3 2 z s 2 1 p i t 3 3 c y 4 f u 4 2
LSCR • The king anchor is an active anchor that is the smallest closest anchor among all active anchors • One king anchor per prefix among known anchors • Forward ALSA from king and HOLD other’s • King Selection Condition (KSO) to select anchor k as king anchor o The anchor advertises that prefix ( vf i =1 ) • k is closest anchor ( rd ik < ¡rd ia ) or • m is at the same distance as other anchors and has smallest ID ( ∀ [ a ] ∈ PAI ij ( rd ik = ¡rd ia ∧ | k | < | a |)) • Distance to a prefix is the distance to king anchor of the prefix
LSCR • Successor-Set Ordering Condition (SOC) to select neighbor n as valid next hop toward prefix j if neighbor n : o is in finite distance toward prefix j ¡ ( d ijn < ∞ ) • is closer to the prefix than router i ( d ijn < d ij ) or • Is in the same distance and has smaller name than the router itself ( d ijn = d ij ∧ | n | < | i |) • SOC prevents permanent routing loops from being created
LSCR: ¡Example q,2 q,2 r,2 q,1 n d v g p,2 p,1 q,1 a r,1 o e x q h r,1 r q,1 b p,2 z s q,1 p i t p,1 c r,1 y f u p,2 r,2 p,2 p,1 • p , q , and r are instances • Valid next hops to nearest anchor
COMPLEXITY ¡ DCR: • o CC = O(EC), SC = O(C) LSR: • o CC = O(CER + lEN), SC = O(RC + E) LSCR: • o CC add = O(C), CC del = O(CER), SC = O(C + E) ü N: number of routers ü E: number of links ü D: number of distinct anchors ü C: number of prefixes ü R: average number of instances ü l: average number of neighbors ü d: network diameter
PERFORMANCE • Simulation Scenario: SCo-Net, NS3 tool o AT&T topology o 210 content objects o 30 anchors o • Initialization: × 10 4 × 10 5 3 1.15 2 1.1 1.75 2.75 Number of AnchorLSAs Number of Operations Anchors Per Prefix 1.05 1.5 2.5 1 1.25 2.25 0.95 1 2 0.9 0.75 NLSR NLSR Discovered 1.75 LSCR LSCR Participated 0.85 0.5 0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 Prefix Instances Prefix Instances Prefix Instances (b) (c) (a)
PERFORMANCE • Simulation Scenario: SCo-Net, NS3 tool o AT&T topology o 210 content objects o 30 anchors o • Computation: × 10 4 × 10 4 5.8 1000 6 5.6 900 5.4 800 5.5 Number of Operations Number of Operations Number of Operations 5.2 700 5 600 5 4.8 500 4.6 400 4.5 4.4 300 4.2 200 4 LSCR NLSR LSCR 4 100 NLSR LSCR NLSR 3.8 0 3.5 0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 Prefix Instances Prefix Instances Prefix Instances (c) (b) (a)
PERFORMANCE • Simulation Scenario: 250 SCo-Net, NS3 tool o 200 Number of AnchorLSAs AT&T topology o 210 content objects 150 o 30 anchors o 100 • Add Prefix: 50 LSCR NLSR 0 0 2 4 6 Prefix Instances (a) 900 800 700 Number of Operations 600 500 400 300 200 LSCR 100 NLSR 0 0 1 2 3 4 5 6 Prefix Instances (b)
Conclusion • LSCR offers: o Less Storage o Less communication • NOC and SOC prevent permanent routing loops from being created • KSO limits forwarded of LSAs and computation
• Thank You! • Any Question?
Recommend
More recommend