An Energy Efficient Leaser Election Algorithm for Mobile Ad Hoc Networks Paolo Cemim, Vinicius De Antoni Instituto de Informatica Universidade Federal do Rio Grande Do Sul Leila Melit, Nadjib Badache 10th International Symposium on Programming and Systems (ISPS), 2011

Introduction A mobile ad hoc network or MANET, is a collection of mobile nodes that can communicate passing over wireless links. o Concurrent and unpredictable topology changes o Dynamic wireless link formation and removal o Network partitioning and disconnections o Limited bandwidth and energy o Highly variable message delay Nodes that are in transmission range can communicate directly o Otherwise they communicate via message relay

Introduction Leader election is a fundamental control problem in both wired and wireless network o However, in MANETS we have:  Link changes  Network splitting into multiple connected components  Merging of multiple connected components each one with its own leader

Introduction The leader election on MANETS is the problem of guaranteeing that: o "any component of the mobile ad hoc network whose topology is static sufficiently long will eventually have exactly one leader" A proposed algorithm is based on Ω failures detector

System Model A distributed system composed of a finite set of n > 1 processes Π = {p 1,...,p n }, that communicates by sending and receiving messages o The network is modeled as a dynamically changing, not necessarily connected, undirected graph All the links considered by the model satisfy: o Integrity: Process q receives a message m from process p at most once, and only if p previoulsy sent m to q o Fair lossy: if p sends infinite messages to q and q is correct, then q receives infinite messages from p o Eventual timely link: There is a time t and a bound d such that each message sent by p is received by q by time t + d

System Model Failures detector o Mechanism that provides information about process failures o Each process has a failure detector monitor  Each monitor maintain a list of processes that it currently suspects to have crashed Failure detector Ω o At every process p and at each time t, the output of the the failure detector Ω is a single process q.  p trusts q to be up at time t. o Ω ensures that eventually all correct process trusts the same process and that this process is correct

System Model A failure detector Ω can be seen as an algorithm for electing a leader: the process trusted by all correct processes is elected Model property o For every component C, there is a correct process l in C and a time t after which for each process p in C, leaderp = l

Leader Election Algorithm Each process p starts the execution by electing itself During its execution it checks if leaderp = p o If so, it sends a message (ALIVE, p) to all every time t Each process p, receiving a message (ALIVE, q) tests whether q is less than or equal to leaderp o If it is equal the process p broadcast the same message o If it is less than, leaderp = q and then broadcast (ALIVE, q) To minimize the number of messages, every process maintain a list of its neighbors and broacast the message to all but the current leader

Leader Election Algorithm Init 1.leaderp <- p; 2.timeoutp <- t; 3.Set timerp to timeoutp; 4.Start task 0 and task 1; Task 0: 1.loop forever 2. if leaderp = p then 3. for each r  Vp do 4. if r> p then send (ALIVE,p) to r every t time; Task 1: 1.upon receotion of message (ALIVE,q) from s such that q <= leaderp do 2. if q < leaderp then leaderp <- q; 3. for each r  Vp do 4. if (r>q) && (r<>s) then send (ALIVE, q) to r; 5. reset timerp to timeoutp; 6.upon experation of timer do: 7. leaderp <- p; 8. timeoutp<- timeoutp + 1 ;

Correctness Lemma 1 o There is a time after which P min permanently satisfies that leader Pmin = P min and sends a message (ALIVE, P min ) every t time Lemma 2 o There is a time after which every message (ALIVE, p) with p != P min dissapears from the system Theorem o There is a time after which each process p in the same component have leader Pmin = P min where P min is the process that has the smallest identifier in that component

Examples

Conclusion of the article Leader highly adaptive to arbitrary and concurrent topological changes Algorithm ensures each connected component has exactly one leader Algorithm is efficient in number of exchanged messages and requires processes to communicate with only a subset of their current neighbors

Conclusion of the group It does not conduct experiments or comparison with related work In some parts of the text, especially in the description of the algorithm, is confusing or difficult comprehension It does not perform the analysis of time complexity or messages exchanged in the algorithm In short, we believe that this is a low-quality paper

Paper Evaluation Motivação e estado-da-arte (2) Problemas a resolver e Modelo (2) Protótipo, Resultados e comparação com trabalhos relacionados (1) Redação e formatação (2)