Analyzing Kleinberg’s (and other) Small-world Models Chip Martel and Van Nguyen Computer Science Department; University of California at Davis 1 Contents Part I: An introduction Background and our initial results Part II: Our new results The tight bound on decentralized routing The diameter bound and extensions An abstract framework for small-world graphs Part III: Future research 2 Our new results For the general k-dimensional lattice model 1. The expected diameter of Kleinbeg’s graph is (log n) 2. The expected length of Kleinberg’s greedy paths is (log2 n). Also, they are this long with constant probability. 3. With some extra local knowledge we can improve the path length to O(log1+1/k n) 3 Background Small-world phenomenon From a popular situation where two completely unacquainted people meet and discover that they are two ends of a very short chain of acquaintances Milgram’s pioneering work (1967): “six degrees of separation between any two Americans” 4 Modeling Small-Worlds Many real settings exhibit small-world properties Motivated models of small-worlds: (Watts-Strogatz, Kleinberg) New Analysis and Algorithms Applications: gossip protocols: Kemper, Kleinberg, and Demers peer-to-peer systems: Malki, Naor, and Ratajczak secure distributed protocols 5 Kleinberg’s Basic setting 6 Kleinberg’s results A decentralized routing problem For nodes s, t with known lattice coordinates, find a short path from s to t. At any step, can only use local information, Kleinberg suggests a simple greedy algorithm and analyzes it: 7 Our Main results For Kleinberg’s small-world setting we Analyze the Diameter for 0 r 2 Give a tight analysis of greedy routing Suggest better routing algorithms A framework for graphs of low diameter. 8 O(log n) Expected Diameter Proof for simple setting: 2D grid with wraparound 4 random links per node, with r=2 Extend to: K-D grids, 1 random link, No wraparound 0r k 9 The diameter bound: Intuition We construct neighbor trees from s and to t: S 0 is the nodes within logn of s in the grid S i is nodes at distance i (random links) from S 0 s S0 10 T-Tree T 0 is the nodes within logn of t in the grid Ti is nodes at distance i (random links) to t T0 T0 11 Subset chains After O(logn) Growth steps S j and Ti are almost surely of size nlogn Thus the trees almost surely connect Similar to Bollobas-Chung approach for a ring + random matching. But new complications since non-uniform distiribution 12 Proving Exponential Growth Growth rate depends on set size and shape We analyze using an artificial experiment 13 Links into or out of a ball Motivation Links to outside Given: subset C , node u, a random link from u. What is the chance for this link to get out of C ? Links into Given: subset C , node u C. What is the chance to have a link to u from outside of C ? Worst shape for C: A ball (with same size) 14 Links into or out of a ball: the facts Bl (u) ={nodes within distance l from u } For a ball with radius n.51 a random link from the center leaves the ball with probability at least .48 With 4 links, expected to hit 4*.48 > 1.9 new 15 nodes from u. S-Tree growth By making the initial set S 0 larger than clogn, a growth step is exponential with probability: 1 n m By choosing c large enough, we can make m large enough so our sets almost surely grow exponentially to size nlogn 16 The t-Tree Ball experiment for t-tree needs some extra care (links are conditioned) Still can show exponential growth Easy to show two (nlogn) size sets of `fresh’ nodes intersect or a link from s-set hits t-set More care on constants leads to a diameter bound of 3logn + o(logn) 17 Diameter Results Thus, for a K-D grid with added link(s) r from u to v proportional to d ( u , v ) The expected diameter is (log n) for 0r k 18 New Diameter Results Thus, for a K-D grid with added link(s) from u to v proportional to d r (u , v ) The expected diameter is (log n) for 0r k New paper: polylog expected diameter for k r 2k Expected diameter is Polynomial for r 2k 19 Analyzing Greedy Routing For r=k (so r=2 for 2D grid), Kleinberg shows greedy routing is O((log2n) . We show this bound is tight, and: With probability greater than ½, Kleinberg’s algorithm uses at least clog2n steps. Fraigniaud et. al also show tight bound, and Suggested by Barriere et. al 1-D result. 20 Proof of the tight bound (ideas) How fast does a step reduce the remaining distance to the destination? We measure the ratio between the distance to t before and after each random trial: We reach t when the product of these ratios is 1 21 Rate of Progress To avoid avoid a product of ratios, we transform to Zv , log of the ratio d(v,t)/d(v’,t) where v’ is the next vertex. Done when sum of Zv totals log(d(s,t)) Show E[Zv] = O(1/logn), so need (log2 n) steps to total log(d(s,t))= logn. 22 An important technical issue: Links to a spherical surface What is the probability to get to a given distance from t ? Let B = {nodes within distance L from t } and SB - its surface Given node v outside B and a random link from v, what is the chance for this link to get to SB? v m L t 23 Extensions Our approach can be easily extended for other lattice-based settings which have: 1. 2. 3. Sufficiency of random links everywhere (to form super node) Rich enough in local links (to form initial S0 and T0 with size (logn)) “Links into or out of a ball” property 24 An abstract framework Motivation: capture the characteristics of KSW model formalize more general classes of SW graphs In the abstract: a base graph, add new random links under a specific distribution Abstract characteristics which result in small diameter and fast greedy routing 25 Part III: Future work The diameter for r=2k (poly-log or polynomial)? Improved algorithms for decentralized routing A routing decision would depend on: the distance from the new node to the destination neighborhood information. Better models for small-world graphs 26