On the Radius of Nonsplit Graphs and Information Dissemination in Dynamic Networks

01/21/2019 ∙ by Matthias Függer, et al. ∙ TU Wien 0

A nonsplit graph is a directed graph where each pair of nodes has a common incoming neighbor. We show that the radius of such graphs is in O( n), where n is the number of nodes. We then generalize the result to products of nonsplit graphs. The analysis of nonsplit graph products has direct implications in the context of distributed systems, where processes operate in rounds and communicate via message passing in each round: communication graphs in several distributed systems naturally relate to nonsplit graphs and the graph product concisely represents relaying messages in such networks. Applying our results, we obtain improved bounds on the dynamic radius of such networks, i.e., the maximum number of rounds until all processes have received a message from a common process, if all processes relay messages in each round. We finally connect the dynamic radius to lower bounds for achieving consensus in dynamic networks.

READ FULL TEXT VIEW PDF
POST COMMENT

Comments

There are no comments yet.

Authors

page 1

page 2

page 3

page 4

This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.

1. Introduction

Consider a distributed system of processes that operate in lock-step synchronous rounds. Let be the set of processes. In a round, each process broadcasts a message and receives messages from a subset of other processes, specified by the directed communication graph whose nodes are the processes and there is an edge in if and only if process receives the message sent by process .

The radius of communication graph is the number of rounds until all processes have (transitively) received a message from a common process. Its value thus poses a lower bound on the number of rounds until information, originating at a single process, can be spread over the entire network. Related applications are from disease spreading and opinion dynamics, where the radius is a lower bound on the rounds it takes to spread a disease or an opinion that originates at a single agent.

Of particular interest in distributed computing are networks that potentially change during execution of an algorithm, be it due to faulty processes, faulty links, mobility of the involved agents, etc.; see, e.g., [10] for a comprehensive overview. We thus generalize the investigation of the radius of a communication graph to the dynamic radius of a sequence of communication graphs . Here, it is assumed that in the above scenario of broadcasting distributed processes, the communication graph for round is . The dynamic radius of the sequence is the number of rounds until all processes have (transitively) received a message from a common process.

1.1. Radius of Nonsplit Digraphs

A nonsplit digraph is a directed graph where each pair of nodes has at least one common incoming neighbor. In this work, we study the radius of nonsplit digraphs: with denoting the length of the shortest path from node to node , the radius is .

In the undirected case, the radius is trivially bounded by the diameter of the graph, which is in the case of nonsplit graphs. Undirected graphs where each pair of nodes has exactly one common neighbor, have been studied by Erdős et al. [3], who showed that they are exactly the windmill graphs, consisting of triangles that share a common node. Thus, their radius is .

As demonstrated by the example in Figure 1 with radius , these bounds do not hold for nonsplit digraphs. We will show the following upper bound:

Theorem 1.

The radius of a nonsplit digraph with nodes is in .

1

2

3

4

5

6
Figure 1. Nonsplit digraph with radius . For example, node and have common incoming neighbor , while nodes and have node as common incoming neighbor.

1.2. Communication over Nonsplit Digraphs

Nonsplit digraphs naturally occur as communication graphs in classical fault-models and as models for dynamic networks.

In fact, several classical fault-models were shown to lead to nonsplit communication graphs [1], among them link failures, as considered in [11], and asynchronous message passing systems with crash failures [10]. Nonsplit digraphs thus represent a convenient abstraction to these classical fault-models. We will show in Section 4 that nonsplit digraphs arising from the classical model of asynchronous messages and crashes have dynamic radius at most .

The study of nonsplit digraphs is also motivated by the study of a central problem in distributed computing: Agreeing on a common value in a distributed system is a problem that lies at the heart of many distributed computing problems, occurs in several flavors, and thus received considerable attention in distributed computing. However, even modest network dynamics already prohibit solvability of exact consensus, where agents have to decide on a single output value that is within the range of the agents’ initial values [11]. For several problems, e.g., distributed control, clock synchronization, load balancing, etc., it is sufficient to asymptotically converge to the same value (asymptotic consensus), or decide on values not too far from each other (approximate consensus). Charron-Bost et al. [1] showed that both problems are solvable efficiently in the case of communication graphs that may vary arbitrarily, but are required to be nonsplit. They further showed that in the more general case where all communication graphs are required to contain a rooted spanning tree, one can in fact simulate nonsplit communication graphs, leading to efficient algorithms for asymptotic and approximate consensus.

Motivated by this work on varying communication graphs, we will show that the following generalization of Theorem 1 holds:

Theorem 2.

The dynamic radius of a network on  nodes whose communication graphs are all nonsplit is .

Traditionally, information dissemination is studied w.r.t. either all-to-all message relay or the time it takes for a fixed process to broadcast its message to everyone [6, 7]. In dynamic networks with nonsplit communication graphs, however, such strong forms of information dissemination are impossible. This can easily be seen by constructing appropriate sequences of star graphs (with self-loops), which are a degenerate case of a nonsplit graph.

One-to-all broadcast of some process, on the other hand, is readily achieved in such networks, which is why we focus on this characteristic here. While it is certainly not as universal as the previously mentioned primitives, it turns out that this type of information dissemination is crucial to the termination time of certain consensus algorithms based on vertex-stable root components [15]. Furthermore, we show the following theorem, relating the dynamic radius and the termination time of arbitrary consensus algorithms:

Theorem 3.

If the dynamic radius of a sequence of communication graphs is , then, in every deterministic consensus algorithm, some process has not terminated before time .

Finally, we note that the dynamic radius is also an upper bound for a single process aggregating the data of all other processes, when we use the alternative interpretation of an edge in a communication graph as a message sent by and received by . Even though this might not be the desired form of data aggregation in a standard setting, in a scenario where the communication is so constrained that aggregation by an a priori selected process is simply unobtainable, such a weak form might still be useful to transmit the collected data to a dedicated sink at regular intervals, for example.

We give a brief overview on related work in the next section.

1.3. Related Work

Information dissemination among an ensemble of participants is a fundamental question that has been studied in a grand variety of settings and flavors (see [4, 6, 7, 8] for various reviews on the topic). While traditional approaches usually assume a static underlying network topology, with the rise of pervasive wireless devices, more recently, focus has shifted to dynamically changing network topologies [9]. A useful way of viewing the distribution of information is to denote the pieces of information that should be shared among the participants as tokens. For instance, the all-to-all token dissemination problem investigates the complete dissemination of initially distributed tokens. This problem was studied in [9] with a focus on bounds for the time complexity of the problem, i.e., how long it takes at least, resp. at most, until tokens have been received by everyone. Here, the participants employed a token-forwarding algorithm mechanism, where tokens are stored and forwarded but not altered.

In the model of [9], it was assumed that the communication graphs are connected and undirected. For this, a lower bound of and an upper bound of for all-to-all token dissemination was established in the case where is unknown to the participants, they have to terminate when the broadcast is finished, and the system is -interval connected, i.e., the communication graphs are completely independent of each other. In contrast, if the communication graphs are connected, directed, and rooted, in the worst case only one of the tokens may ever be delivered to all participants. This can be seen, for example, when considering a dynamic graph that produces the same line graph for every round. We note that this example also provides a trivial lower bound of rounds until one token is received by everyone for the first time. As far as we are aware, the best such lower bound was established in [13, Theorem 4.3] to be rounds. Studying directed graphs is desirable as they represent a weaker, more general model and wireless communication is often inherently directed, for example due to localized fading or interference phenomena [12, 5] such as the capture effect or near-far problems [14].

In [2], it was shown that the dynamic radius of a sequence of arbitrary nonsplit communication graphs is . Later, it was shown in [1] that the product of any rooted communication graphs is nonsplit. Put together, this means that the dynamic radius of a sequence of arbitrary rooted graphs is . More recently, [13] provided an alternative proof for this fact that does not rely on the reduction to nonsplit graphs but instead uses a notion of influence sets similar to [9, Lemma 3.2.(b)]. In addition to this, [13] provided linear bounds in sequences of rooted trees with a constant number of leaves or inner nodes, established a dependency on the size of certain subtrees in sequences of rooted trees where the root remains the same, and investigated sequences of undirected trees.

2. Model and Definitions

We start with some definitions motivated by the study of information dissemination within a distributed system of processes that operate in discrete, lock-step synchronous communication rounds. Starting with information being available only locally to each process, processes broadcast and receive information tokens in every round. We are interested in the earliest round where all processes have received an information token from a common process.

Clearly, the dissemination dynamics depends on the dynamics of the underlying network. For this purpose we define: A communication graph on nodes is a directed graph with self-loops and the set of nodes . For , let denote the set of in-neighbors of in and denote its set of out-neighbors. Intuitively, communication graphs encode successful message reception within a round: an edge from to states that received the message broadcast by in this round.

A node is called a broadcaster in  if it has an edge to all nodes, i.e., .

A communication graph is nonsplit if every pair of nodes has a common incoming neighbor, i.e.,

Given two communication graphs and on nodes, define their product graph as where

The empty product is equal to the communication graph which contains the self-loops for all nodes  and no other edges. The graph product we use here is motivated by information dissemination within distributed systems of processes that continuously relay information tokens that they received: if received ’s information token in a round, and received ’s information token in the next round, then received ’s information token in the macro-round formed by these two successive rounds.

Motivated by modeling communication networks that potentially change in each round, we call each infinite sequence of communication graphs on  nodes a communication pattern on  nodes. For every node , define the broadcast time of node  in  as the minimum  such that  is a broadcaster in the product of the first  communication graphs of . If no such  exists, then . The dynamic radius of  is the minimal broadcast time of its nodes, i.e., . Note that is the earliest time, in terms of rounds, until that all nodes have received an information token from a common node, given that the communication pattern is .

A network on nodes is a nonempty set of communication patterns on  nodes; modeling potential uncertainty in a dynamic communication network. A network’s dynamic radius is defined as the supremum over all dynamic radii of its communication patterns, capturing the worst-case of information dissemination within this network.

3. The Dynamic Radius of Nonsplit Networks

In this section we show an upper bound on the dynamic radius of nonsplit networks.

During the section, let be a communication pattern on  nodes in which every communication graph  is nonsplit.

In order to prove an upper bound on the dynamic radius of , we will prove the existence of a relatively small set of nodes that ”infects” all other nodes within only rounds. Iteratively going back in time, it remains to show that any such set is itself ”infected” by successively smaller and smaller sets within rounds, until we reach a single node. It follows that this single node has ”infected” all nodes with its information token after rounds.

Note that the strategy to follow ”infection” back in time rather than consider the evolution of infected sets over time is essential in our proofs: it may very well be that a certain set of infected nodes cannot infect other nodes from some time on, since it only has incoming edges from nodes not in the set in all successive communication graphs. Going back in time prevents us to run into such dead-ends of infection.

For that purpose we define: Let be sets of nodes. We say that  covers  in communication graph if for every there is some that has an edge to , i.e., .

Now let . We say that at time  covers  at time  if  covers  in the product graph .

Note that  at time  covers  at time  for all sets and all , by definition of the empty product as the digraph with only self-loops.

We first show that the notion of covering is transitive:

Lemma 1.

Let and let . If  at time  covers  at time , and  at time  covers  at time , then at time  covers  at time .

Proof.

By definition, for all there is some such that is an edge of the product graph . Also, for all there is some such that is an edge of the product graph .

But, by the associativity of the graph product, this means that for all there exists some such that is an edge in the product graph

That is, at time  covers  at time . ∎

Lemma 2.

For all we have .

Proof.

We have if . Now, noting that the inequality is equivalent to whenever  is an integer concludes the proof. ∎

Lemma 3.

Let  and  be positive integers such that . Then .

Proof.

W.l.o.g., let . We distinguish between two cases for positive integer :

(i) If then , and we immediately obtain the lemma from and .

(ii) Otherwise, . From we deduce . This implies

We are hence done if we can show . But this is just Lemma 2 with . ∎

Lemma 4.

Let  and  be positive integers such that . Then there exist positive integers such that and for all .

Proof.

Let with and be the integer division of  by . Set and .

By Lemma 2, we have

for all . ∎

We continue with the following generalization of a result by Charron-Bost and Schiper [2]. In particular (), it shows that any set of nodes can be ”infected” by a single node, such that the set of infected nodes grows exponentially in size per round.

Lemma 5.

Let be nonempty and be a positive integer. If , then there exists some with such that  at time  covers  at time .

Proof.

Using Lemma 4, we can assume without loss of generality that . We proceed by induction on .

Base case: If , i.e., , then and the statement is trivially true since we can choose .

Inductive step: Now let . Let such that . Using Lemma 3, we see that for . By the induction hypothesis, there hence exist nodes  and  that at time cover  and , respectively. But now, using the nonsplit property of communication graph , we see that there exists a node  that covers in . An application of Lemma 1 concludes the proof. ∎

Note that Lemma 5, by choosing and , provides an upper bound on the dynamic radius of . To show an upper bound of , we will apply this lemma only for the early infection phase of rounds, and use a different technique, by the next two lemmas, for the late phase.

Lemma 6.

Let  and  be finite sets with , , and . If , then there exists some such that .

Proof.

By the pigeonhole principle, we get the existence of some with

(1)

Write and . Since for all , we have

(2)

Combining (1) and (2), we get

(3)

where we used . This concludes the proof. ∎

The next lemma shows that during late infection, nodes are infected faster than than exponential as provided by Lemma 5 for the early phase:

Lemma 7.

There exists some  such that for all there exists a set of at most nodes that at time  covers the set  of all nodes at time .

Proof.

For every set of nodes, let be a node that at time  covers  at time , which exists by Lemma 5.

We recursively define the following sequence of nodes , and sets of nodes , :

  • For , we choose  such that , which exists by Lemma 6, and .

Note that, setting , we have . Hence the set at time  covers all nodes at time . Noting concludes the proof. ∎

We are now ready to combine Lemma 5 for the early phase and Lemma 7 for the late phase to prove the main result of this section.

Theorem 4.

The dynamic radius of a network on  nodes whose communication graphs are all nonsplit is .

Proof.

Let where  is the constant from Lemma 7.

By Lemma 7, there is a set  of nodes with that at time  covers all nodes at time . By Lemma 5, a single node at time  covers  at time .

Combining both results via Lemma 1 shows that a single node at time  covers all nodes at time . ∎

4. Nonsplit Networks from Asynchronous Rounds

We now show that in an important special case of nonsplit networks, namely those evolving from distributed algorithms that establish a round structure over asynchronous message passing in the presence of crashes, the dynamic radius is at most .

In the classic asynchronous message passing model with crashes, it is assumed that all messages sent have an unbounded but finite delay until they are delivered. Furthermore, processes do not operate in lock-step but may perform their computations at arbitrary times relative to each other. In addition, some processes may be faulty in the sense that they are prone to crashes, i.e., they may seize to perform computations at an arbitrary point in time.

This means that in a system where up to processes may be faulty, in order to progress in a distributed algorithm, a process may wait until it received a message from different processes but no more: If a process waits for a message from different processes, but there were in fact crashes, this process will wait forever. For this reason, algorithms for this asynchronous model often employ the concept of an asynchronous round, sometimes realized as a local round counter variable , which is held by each process and appended to every message. A process increments only if it received a message containing a round counter from different processes.

One may now ask how fast information tokens can spread in such a distributed system, again, if processes repeatedly receive and forward information tokens.

For that purpose we consider the network whose communication patterns are induced by processes communicating in asynchronous rounds. When deriving the communication graph of such an asynchronous round, we get a digraph where each process has at least incoming neighbors. In this sense, the round communication graph represents, for each process , a set of processes that managed to send a message to , containing a round counter , before they crashed. It is important to note that an edge in represents that received a message from that contained a round counter , but not necessarily .

Below, we study the case where , i.e., a majority of the processes is correct. This implies that the sets of incoming neighbors of any two processes in a communication graph have a non-empty intersection, which means that the communication graph is nonsplit. In fact, if , then the network can be disconnected into two disjoint sets of processes that do not receive messages from the other until termination of the algorithm. Below, we establish a constant upper bound on the dynamic radius of this important class of nonsplit graphs.

Theorem 5.

Let , , and be a sequence of communication graphs with for all and all . The dynamic radius of is at most .

Proof.

To show the bound on the radius we prove that there exists a node , that will be the center that realizes the dynamic radius, i.e.,

(4)

Equation (4) now follows from

(5)

by the following arguments: Equation (5) states that the information at has been transmitted to at least nodes. By assumption for all . Thus each must have an incoming neighbor in digraph such that ; equation (4) follows.

It remains to show (5). Suppose that the equation does not hold, i.e.,

(6)

By assumption on digraph , we have

(7)

Denoting by the function that is if its argument is true, and otherwise, we may rewrite

and, using (6),

Together with (7), we have ; a contradiction to the assumption that . ∎

5. A Lower Bound for Consensus in Dynamic Networks

Subsequently, we show that the dynamic radius of a network presents a lower bound on the time complexity of a consensus algorithm for this network.

Let be a set of processes that operate in lock-step synchronous rounds delimited by times where, by convention, round happens between time and time . Each round consists of a phase of communication, followed by a phase of local computation. Like in the previous sections, a communication pattern defines, for each round, which messages reach their destination.

In the (exact) consensus problem, every node starts with an input value from an arbitrary domain and holds a unique write-once variable , initialized to , where denotes a special symbol s.t. . Since we are concerned with an impossibility result here, we may restrict ourselves without loss of generality to the binary consensus problem, i.e., the case where . An execution of a deterministic consensus algorithm is a sequence of state transitions according to the algorithm and determined by the input assignment and the communication pattern. An algorithm solves consensus if it satisfies in all of its executions:

  • (Termination) Eventually for every , .

  • (Validity) If then for some .

  • (Agreement) For every , if and then .

Theorem 6.

If the dynamic radius of the network is , then, in every deterministic consensus algorithm, some process has not terminated before time .

Proof.

Let be a communication pattern with dynamic radius , which occurs in the network by assumption. Suppose, in some deterministic consensus algorithm , all have terminated at time in every execution based on . Let be the input assignment where for all and be the input assignment where for all . By validity, when running under and starting from , all have by time and when starting from , they have . Thus, there are input assignments that differ only in the input assignment of a single process and, for all , at time , when applying under when starting from and when starting from . Since there is no broadcaster in before round , there is some process that did not receive a (transitive) message from and thus is in the same state in both executions. Therefore, decides on the same value in both executions, which is a contradiction and concludes the proof. ∎

6. Conclusion

In this paper, we found that nonsplit networks are a convenient abstraction that arises naturally when considering information dissemination in a variety of dynamic network settings. Since classic information dissemination problems are trivially impossible in these nonsplit dynamic networks, it made sense to study the more relaxed dynamic radius here. As we showed in Theorem 6, this is an important characteristic with respect to the impossibility of exact consensus. For our main technical contribution, we proved a new upper bound in Theorem 4, which shows that the dynamic radius of nonsplit networks is in . This is an exponential improvement of the previously known upper bound of .

In Section 4, we showed an upper bound of asynchronous rounds for the dynamic radius in the asynchronous message passing model with crash failures. Thus, in this important class of nonsplit networks, information dissemination is considerably faster than what is currently known for the general case.

Combining our Theorem 4 with the result from [1] that established a simulation of nonsplit networks in rooted networks, i.e., networks where every communication graph contains a rooted spanning tree, yields an improvement of the previously known upper bound for the dynamic radius of rooted dynamic networks from to :

Theorem 7.

The dynamic radius of a dynamic networks whose communication graphs are rooted is .

While this is another hint at the usefulness of the nonsplit abstraction for dynamic networks, the tightness of this bound remains an open question.

References

  • [1] Bernadette Charron-Bost, Matthias Függer, and Thomas Nowak. Approximate consensus in highly dynamic networks: The role of averaging algorithms. In Magnús M. Halldòrsson, Kazuo Iwama, Naoki Kobayashi, and Bettina Speckmann, editors, Automata, Languages, and Programming, volume 9135 of Lecture Notes in Computer Science, pages 528–539. Springer Berlin Heidelberg, 2015.
  • [2] Bernadette Charron-Bost and André Schiper. The Heard-Of model: computing in distributed systems with benign faults. Distributed Computing, 22(1):49–71, April 2009.
  • [3] Paul Erdös, Alfréd Rényi, and Vera T. Sós. On a problem of graph theory. Studia Sci. Math. Hungar., 1:215–235, 1966.
  • [4] Pierre Fraigniaud and Emmanuel Lazard. Methods and problems of communication in usual networks. Discrete Applied Mathematics, 53(1):79 – 133, 1994.
  • [5] Alois Goiser, Samar Khattab, Gerhard Fassl, and Ulrich Schmid. A new robust interference reduction scheme for low complexity direct-sequence spread-spectrum receivers: Performance. In Proceedings 3rd International IEEE Conference on Communication Theory, Reliability, and Quality of Service (CTRQ’10), pages 15–21, Athens, Greece, June 2010.
  • [6] H. A. Harutyunyan, A. L. Liestman, J. Peters, and D. Richards. Broadcasting and gossiping). In J. Gross and P. Zhang, editors, The Handbook of Graph Theory, pages 1477–1494. Chapman and Hall/CRC, 2013.
  • [7] Sandra M. Hedetniemi, Stephen T. Hedetniemi, and Arthur L. Liestman. A survey of gossiping and broadcasting in communication networks. Networks, 18(4):319–349, 1988.
  • [8] Juraj Hromkovič, Ralf Klasing, Burkhard Monien, and Regine Peine. Dissemination of information in interconnection networks (broadcasting & gossiping). In Ding-Zhu Du and D. Frank Hsu, editors, Combinatorial Network Theory, pages 125–212. Springer US, Boston, MA, 1996.
  • [9] Fabian Kuhn, Nancy A. Lynch, and Rotem Oshman. Distributed computation in dynamic networks. In STOC, pages 513–522, 2010.
  • [10] Nancy A Lynch. Distributed algorithms. Elsevier, 1996.
  • [11] Nicola Santoro and Peter Widmayer. Time is not a healer. In B. Monien and R. Cori, editors, 6th Symposium on Theoretical Aspects of Computer Science, volume 349 of LNCS, pages 304–313. Springer, Heidelberg, 1989.
  • [12] Udo Schilcher, Christian Bettstetter, and Günther Brandner. Temporal correlation of interference in wireless networks with rayleigh block fading. IEEE Transactions on Mobile Computing, 11(12):2109–2120, 2012.
  • [13] M. Schwarz, M. Zeiner, and U. Schmid. Linear-Time Data Dissemination in Dynamic Networks. ArXiv e-prints: https://arxiv.org/abs/1701.06800, January 2017.
  • [14] C. Ware, J. Judge, J. Chicharo, and E. Dutkiewicz. Unfairness and capture behaviour in 802.11 adhoc networks. In 2000 IEEE International Conference on Communications. ICC 2000. Global Convergence Through Communications., 2000.
  • [15] Kyrill Winkler, Manfred Schwarz, and Ulrich Schmid. Consensus in Rooted Dynamic Networks with Short-Lived Stability. arXiv e-prints, page arXiv:1602.05852, February 2016.