We present Greedy Perimeter Stateless Routing (GPSR), a novel routing protocol for wireless datagram networks that uses the positions of routers and a packet's destination to make packet forwarding decisions. GPSR makes greedy forwarding decisions using only information about a router's immediate neighbors in the network topology. When a packet reaches a region where greedy forwarding is impossible, the algorithm recovers by routing around the perimeter of the region. By keeping state only about the local topology, GPSR scales better in per-router state than shortest-path and ad-hoc routing protocols as the number of network destinations increases. Under mobility's frequent topology changes, GPSR can use local topology information to find correct new routes quickly. We describe the GPSR protocol, and use extensive simulation of mobile wireless networks to compare its performance with that of Dynamic Source Routing. Our simulations demonstrate GPSR's scalability on densely deployed wireless networks.

/pdf/gpsr-greedy-perimeter-stateless-routing-for-wireless-59gz21qf8y.pdf

GPSR: greedy perimeter stateless routing for wireless networks

https://www.cs.montana.edu/courses/csci566/Ref/CR-Survey.pdf

NeXt generation/dynamic spectrum access/cognitive radio wireless networks: a survey

Advances in processor, memory and radio technology will enable small and cheap nodes capable of sensing, communication and computation. Networks of such nodes can coordinate to perform distributed sensing of environmental phenomena. In this paper, we explore the directed diffusion paradigm for such coordination. Directed diffusion is datacentric in that all communication is for named data. All nodes in a directed diffusion-based network are application-aware. This enables diffusion to achieve energy savings by selecting empirically good paths and by caching and processing data in-network. We explore and evaluate the use of directed diffusion for a simple remote-surveillance sensor network.

/pdf/directed-diffusion-a-scalable-and-robust-communication-4tsx5vxsas.pdf

Directed diffusion: a scalable and robust communication paradigm for sensor networks

"A Survey of Mobility Models for Ad Hoc Network Research," Wireless Comm. & Mobile Computing (WCMC) : Special issue on Mobile Ad Hoc Networking : Research

In the performance evaluation of a protocol for an ad hoc network, the protocol should be tested under realistic conditions including, but not limited to, a sensible transmission range, limited buffer space for the storage of messages, representative data traffic models, and realistic movements of the mobile users (i.e., a mobility model). This paper is a survey of mobility models that are used in the simulations of ad hoc networks. We describe several mobility models that represent mobile nodes whose movements are independent of each other (i.e., entity mobility models) and several mobility models that represent mobile nodes whose movements are dependent on each other (i.e., group mobility models). The goal of this paper is to present a number of mobility models in order to offer researchers more informed choices when they are deciding upon a mobility model to use in their performance evaluations. Lastly, we present simulation results that illustrate the importance of choosing a mobility model in the simulation of an ad hoc network protocol. Specifically, we illustrate how the performance results of an ad hoc network protocol drastically change as a result of changing the mobility model simulated.

/pdf/a-survey-of-mobility-models-for-ad-hoc-network-research-ygkxat400x.pdf

A survey of mobility models for ad hoc network research

1. INTRODUCTION N ! .O! .P PXJLY: [Z ,,\ , O P*P A 'T[$]P_^#`a`b`dc' ; e-eK P f .,-g , Z+ h$' Ui j , & .& , & P ,-O A , &%kL lTm$'& !P kLPFno, QP Tm$' A e-eXz k3,\(A 7RF& ,\(A P 7 lTm$'Aj $'Ak3 P S+ Tm$ U7 A$]O! ?$'n L$Q ,-O O P ,-&0

/pdf/handling-mac-layer-misbehavior-in-wireless-networks-331d31djag.pdf

Handling MAC Layer Misbehavior in Wireless Networks

In wireless networks, the collision cost is much higher than wired networks since a station cannot detect collision until a transmission is over and the expected acknowledgment does not come back. Therefore, more efficient contention resolution algorithms are desired for wireless networks to reduce the collision probability among contending stations. IEEE 802.11 defines a distributed random access MAC protocol named DCF (Distributed Coordination Function), which uses binary exponential backoff (BEB) algorithm to resolve channel contention. In DCF, a station wanting to access the channel generates a random backoff counter uniformly distributed over the interval [0, CW] (CW represents contention window). This backoff counter corresponds to the number of idle slots this station has to wait before its transmission. The contention window, CW, is exponentially increased by a factor of 2 each time when a collision happens, and reset to the minimum value upon successful transmission. An appropriate choice of CW can optimize the performance of 802.11. However, the optimum value of CW changes with the network size and 802.11 operates far from this optimum point. In particular, in a heavily contended network, the collision probability increases significantly, degrading the performance of 802.11. To improve the performance of IEEE 802.11, [1] proposes that each station continuously observes channel activities and estimates the number of contending stations, hence, dynamically tune the value of CW to the optimum point. [2] also proposes a fast collision resolution algorithm, in which the winning station occupies the channel for a certain period of time to reduce the collision probability. However, both of these two schemes are designed for wireless LANs, where all stations are within each other’s transmission range. This paper proposes a MAC protocol, named DSCR (Dual Stage Contention Resolution), uses “pipelined” two stage contention resolution algorithm to reduce collision probability and achieve better channel utilization than 802.11 in both wireless LANs and ad hoc networks. In our prior work, we developed two explicit pipelining schemes, which use a separate control channel (either to transmit RTS/CTS or to send busy tone) to pipeline contention resolution with data transmission [4]. In contrast, DSCR performs implicit pipelining in that no separate control channel is required.

DSCR: a wireless MAC protocol using implicit pipelining

In this paper, we propose to reduce the effect of rate-independent MAC overheads in random access protocols by partitioning the transmission channel spectrum into a narrow channel and a wide channel. The narrow channel is used for transmitting the short packets (approximately 100 bytes long) and the wide channel is used for transmitting the longer packets. We intend to use multiple radios, one each for the different channel partitions. Narrow width channels have a reduced capacity, which lowers the maximum transmission rate achievable on these channels. As a result, the channel wastage due to the rate-independent MAC overheads can be reduced. We propose a protocol called WiSP (channel Width Selection based on Packet size) to estimate the appropriate channel widths depending on the relative traffic load involving short and long packets in the network. We evaluate our protocol using extensive simulations and demonstrate its effectiveness in achieving higher throughputs. We propose our algorithm to complement the frame aggregation (an existing approach that aggregates multiple packets to be sent in a single transmit opportunity) technique. We show that there are scenarios during which the frame aggregation can perform poorly, and show that our proposed algorithm can provide a good performance even in those situations when used along with frame aggregation.

WiSP: A protocol for overcoming MAC overheads using packet size dependent channel widths

Causally consistent distributed storage systems have received significant attention recently due to the potential for providing high throughput and causality guarantees. {\em Global stabilization} is a technique established for ensuring causal consistency in distributed storage systems, adopted by the previous work such as GentleRain \cite{Du2014GentleRainCA} and Cure \cite{akkoorath2016cure}. It allows the client to read consistently without explicit dependency checking, thus enables low latency and high throughput. However, most of the previous designs assume {\em full replication}, where each data center stores a full copy of data. In this paper, we extend global stabilization to general {\em partial replication}, where each server can store an arbitrary subset of the data, and the clients are allowed to communicate with any subset of the servers. We propose an algorithm that implements causally consistency partially replicated distributed storage using global stabilization. We prove the correctness of our algorithm and provide an efficient implementation. We also discuss several optimizations that can improve the performance of our algorithm.

Global Stabilization for Causally Consistent Partial Replication

This paper considers the Byzantine fault-tolerance problem in distributed stochastic gradient descent (D-SGD) method – a popular algorithm for distributed multi-agent machine learning. In this problem, each agent samples data points independently from a certain data-generating distribution. In the fault-free case, the D-SGD method allows all the agents to learn a mathematical model best fitting the data collectively sampled by all agents. We consider the case when a fraction of agents may be Byzantine faulty. Such faulty agents may not follow a prescribed algorithm correctly, and may render traditional D-SGD method ineffective by sharing arbitrary incorrect stochastic gradients. We propose a norm-based gradient-filter, named comparative gradient elimination (CGE), that robustifies the D-SGD method against Byzantine agents. We show that the CGE gradient-filter guarantees fault-tolerance against a bounded fraction of Byzantine agents under standard stochastic assumptions, and is computationally simpler compared to many existing gradient-filters such as multi-KRUM, geometric median-of-means, and the spectral filters. We empirically show, by simulating distributed learning on neural networks, that the fault-tolerance of CGE is comparable to that of existing gradient-filters. We also empirically show that exponential averaging of stochastic gradients improves the fault-tolerance of a generic gradient-filter.

Nitin H. Vaidya

Papers

Handling MAC Layer Misbehavior in Wireless Networks

DSCR: a wireless MAC protocol using implicit pipelining

WiSP: A protocol for overcoming MAC overheads using packet size dependent channel widths

Global Stabilization for Causally Consistent Partial Replication

Byzantine Fault-Tolerant Distributed Machine Learning with Norm-Based Comparative Gradient Elimination