PaperlayerView exact revision on arXiv

Source-linked AI summary

Data Collection and Wireless Communication in Internet of Things (IoT) Using Economic Analysis and Pricing Models: A Survey

Nguyen Cong Luong, Dinh Thai Hoang, Ping Wang, Dusit Niyato, Dong In Kim, Zhu Han

arXiv:1608.03475v1cs.GTcs.CY

TL;DR

IoT data collection and communication require adaptive decisions under constrained resources, dynamic environments, and heterogeneous participants. This paper surveys economic and pricing models across WSNs, crowdsensing, and M2M communication, reviewing their applications, advantages, disadvantages, and open research directions. The survey organizes these approaches across major sensing, communication, allocation, coverage, security, and participation problems, including reported improvements in traffic prediction and energy savings.

  • Problem

    IoT systems need adaptive and robust approaches for data collection, communication, resource optimization, coverage, task allocation, and security under constrained and dynamic conditions.

  • Method

    The paper provides a state-of-the-art survey and comparison of economic and pricing models applied to WSNs, participatory and crowdsensing networks, and M2M communication.

  • Results

    The survey covers economic and pricing applications across IoT sensing and communication problems, including crowdsensing incentives, utility optimization, target tracking, and M2M resource allocation.

  • Takeaways & Limitations

    Economic and pricing models provide a framework for adaptive IoT algorithms and protocols, while the survey identifies open issues and future research directions.

Abstract

from arXiv · show

This paper provides a state-of-the-art literature review on economic analysis and pricing models for data collection and wireless communication in Internet of Things (IoT). Wireless Sensor Networks (WSNs) are the main component of IoT which collect data from the environment and transmit the data to the sink nodes. For long service time and low maintenance cost, WSNs require adaptive and robust designs to address many issues, e.g., data collection, topology formation, packet forwarding, resource and power optimization, coverage optimization, efficient task allocation, and security. For these issues, sensors have to make optimal decisions from current capabilities and available strategies to achieve desirable goals. This paper reviews numerous applications of the economic and pricing models, known as intelligent rational decision-making methods, to develop adaptive algorithms and protocols for WSNs. Besides, we survey a variety of pricing strategies in providing incentives for phone users in crowdsensing applications to contribute their sensing data. Furthermore, we consider the use of some pricing models in Machine-to-Machine (M2M) communication. Finally, we highlight some important open research issues as well as future research directions of applying economic and pricing models to IoT.

I. INTRODUCTION

The paper surveys economic and pricing models for adaptive IoT data collection and communication, focusing on WSN challenges, participating entities, and sensing incentives. It organizes prior work across IoT architectures, services, and major WSN issues.

  • IoT connects billions of smart devices that sense, process, and transmit data with minimal human intervention.
  • WSN sensors must make optimal decisions under constrained resources and dynamic environments while supporting IoT services.
  • Economic and pricing approaches address interactions among self-interested entities with different objectives and constraints.
  • Pricing and payment strategies can incentivize crowdsensing participation and improve sensing accuracy, coverage, and timeliness.
  • Auctions can select sensors with greater remaining resources, balancing network lifetime against required data quality and reducing redundancy.
  • The survey classifies work around data exchange and topology, resource and power allocation, sensing coverage, security, related IoT problems, and M2M resource allocation.

C. Resources and Services of IoT

IoT resources and services—including sensing data, power, cloud services, bandwidth, information, and location services—can be traded or priced. The paper introduces economic and pricing approaches for managing these resources and their associated value.

  • Resource management is a key challenge because IoT is heterogeneous, large-scale, and composed of multiple resources and services.
  • Sensing data can be traded and priced to optimize the profits of device owners and service providers.
  • IoT markets include power, cloud storage and computation, spectrum and bandwidth, data and information services, and location-based services.
  • Economic and pricing approaches are categorized as economic-concept, game-theory and auction, or optimization-based pricing.
  • Cost-based pricing sets selling price as p = C×(1+m), incorporating fixed and variable costs plus a markup.
  • Consumer perceived value pricing estimates demand through willingness and affordability, considering utility, credibility, motivations, and context.

3) Supply and demand model:

The paper presents supply-and-demand and smart-data-pricing concepts for valuing IoT sensing data and managing scarce network resources. It also describes option pricing as an approach for investment, reservation, and scheduling decisions.

  • Market equilibrium balances supply and demand for network resources and sensing data while supporting information quality and flexible spectrum sharing.
  • Smart Data Pricing uses time-dependent and usage-based charges to reduce congestion and improve network resource efficiency.
  • Time-dependent pricing varies prices across periods to diffuse demand, reduce peaks, and fill valley periods.
  • Usage-based pricing charges according to access rate or resource usage instead of relying on static flat prices.
  • Option pricing: The Black-Scholes model values options using current expected cash-flow value, exercise price, expiration time, and risk-free interest rate.
  • Option pricing: In IoT, real option theory has been applied to investment evaluation, M2M resource reservation, and sensor task scheduling.

B. Game Theory and Auction Based Pricing

Game theory and auctions model strategic interactions among sensing-data buyers, sellers, and service providers. Non-cooperative and Stackelberg formulations characterize pricing decisions, reactions, and payoff outcomes.

  • Game-theoretic pricing treats buyers, sellers, or service providers as players whose strategies affect one another’s payoffs.
  • Non-cooperative game: A Nash equilibrium occurs when no seller can improve its payoff by unilaterally changing strategy, but existence or uniqueness is not guaranteed.
  • Non-cooperative game: Non-cooperative game theory has been applied to resource allocation, spectrum sharing, cloud-provider profits, and sensing coverage optimization.
  • Stackelberg game: Stackelberg games address settings where strategies are announced sequentially because simultaneous mutual knowledge may not hold in real markets.
  • Stackelberg game: A Stackelberg strategy is optimal for the leader when the follower responds with an optimal strategy.
  • Stackelberg game: The leader obtains at least as good a payoff as under the corresponding Nash solution in the described Stackelberg setting.

3) Bargaining game:

This section introduces bargaining and auction-based pricing mechanisms for sensing data and wireless-resource exchange. It distinguishes auction formats by participant roles and explains their pricing and equilibrium properties.

  • Bargaining game: Bargaining models a seller and buyer choosing offers that maximize their expected profits until a mutually acceptable sensing-data price is reached.The resulting best-response pair forms a Nash bargaining equilibrium when neither player can improve expected profit unilaterally.
  • Auction: Auctions define explicit rules for allocating items such as sensing data or bandwidth and determining participant prices.Participants include buyers or bidders, sellers, and an auctioneer who controls the auction and announces the winner.
  • Sealed-bid auctions: In sealed-bid auctions, buyers submit bids simultaneously without observing or revising bids, unlike open-cry auctions.First-price winners pay their highest submitted price, whereas second-price winners pay the second-highest bid.
  • Auction properties: VCG auctions charge a winner for the social value lost when it receives an item, and their outcome is a Bayes-Nash equilibrium.The survey notes that sealed-bid auctions are more commonly used because of their simplicity and privacy guarantee.
  • Auction types: Forward auctions have buyers bid upward, reverse auctions have sellers compete through decreasing asks, and double auctions match simultaneous bids and asks.A double auction sets a clearing price, typically p = (p_i + a_j)/2, between a buyer’s bid and seller’s ask.

5) Posted price mechanism:

This section describes posted prices and utility-maximization formulations for allocating IoT resources. It also states the concavity assumption required for a unique optimal allocation and price vector.

  • Posted price mechanism: Posted-price mechanisms give sequentially arriving sellers take-it-or-leave-it offers, avoiding the complexity of soliciting bids in online procurement markets.A seller accepts when its actual cost is below the posted offer.
  • Utility maximization: Network Utility Maximization allocates rates and charges buyers according to utility functions, resource routes, capacities, and per-unit prices.The buyer’s payment is p_s = r_sλ_s, where r_s is allocated rate and λ_s is the charge per unit.
  • Utility maximization: Under increasing, strictly concave, continuously differentiable utilities, the optimization problems have a unique optimal rate-allocation vector and resource-price vector.The vector r* is the unique optimal allocating rate vector, while λ* is the current optimal resource-price vector.
  • Applications: The utility-maximization scheme is also applied to bandwidth supply-demand balance and congestion or contention control.
  • Limitation: For hybrid services with inelastic flows, utilities may make pricing and resource allocation non-convex, so the NUM framework is no longer appropriate.

2) Multi-objective knapsack problem:

This section presents knapsack-based sensor selection under energy constraints and surveys pricing applications across WSN data collection, routing, transmission, and congestion control.

  • Multi-objective knapsack problem: The knapsack formulation selects a subset of active sensors that maximizes aggregate utility subject to a specified energy budget.The budget represents task energy requirements, while each sensor’s weight represents its participation energy cost.
  • Multi-objective knapsack problem: Binary variable x_i indicates whether sensor i participates, while α and β balance application-specific priorities.Although knapsack optimization is NP-hard, dynamic programming solves it optimally in pseudo-polynomial time.
  • Applications: Knapsack models are also used for sensing-data collection and IoT-aware placement algorithms.
  • Data aggregation and routing: Pricing models support WSN data aggregation, opportunistic transmission, relay selection, and congestion management under resource constraints.Data aggregation reduces redundancy and transmission energy, while pricing helps select efficient channels and forwarding relays.
  • Pricing applications: A sealed-bid reverse auction can select high-remaining-energy sensors, whereas value-based pricing sets data prices according to requester requirements such as quality, delay, and packet lifetime.The value-based price is determined in two stages by the sink and an access point.
  • Data aggregation and routing: For damaged networks with up to 30% damage, a hop-by-hop minimum-cost route increased delivered packets while reducing delay more than DSDV; beyond that damage, the difference was slight.

2) Data aggregation and routing in participatory sensing and crowdsensing networks:

Economic and pricing models support participatory and crowdsensing data aggregation by balancing incentive cost, participation, data quality, fairness, and service-provider constraints. The surveyed mechanisms include auctions, credit schemes, posted prices, value-based pricing, and multi-objective user selection.

  • Motivation: Participatory and crowdsensing systems use economic incentives because users consume energy and bandwidth when contributing sensing data.Reviewed approaches target participation while addressing data-aggregation costs and sensing quality.
  • Auction-based mechanisms: RADP-VPC reduces incentive cost by rewarding previous-round losers with virtual participation credit, stabilizing competition across auction rounds.The virtual credit gives losers more opportunities to win in subsequent rounds.
  • Auction-based mechanisms: Adding participant recruitment to RADP-VPC suppresses auction prices by helping dropped users reassess their returns and potentially rejoin later rounds.The mechanism broadcasts the previous winners’ payments as reference information for dropped users.
  • Credit-based mechanisms: Exponential smoothing predicts expected credits from recent payment histories, bringing total allocated credit within ±0.05 of total user requirements.The method cannot serve users without previous payments.
  • Quality and allocation: Data-quality-aware and multi-objective mechanisms address limits of price-only selection by trading payment against quality, QoS, fairness, or participation.The multi-objective Knapsack approach outperforms auction strategies in total data quality relative to total payment, while other schemes use quality, timeliness, coupons, or contribution-based quotas.

B. Opportunistic Transmission and Neighbor Discovery

Opportunistic transmission uses cost-based pricing and neighbor discovery to route sensing data through lower-cost short-range links or selected relays. The surveyed approaches reduce transmission costs and energy use, while leaving threshold-price design and joint aggregation unresolved.

  • Motivation: Cellular capacity and 3G/4G costs motivate opportunistic networking to reduce transmission costs, energy consumption, and traffic.The approach forwards data through nearby users or WiFi routers when appropriate.
  • Cost-based opportunistic transmission: Cost-based pricing lets a source compare relay costs and select the neighbor with the minimum input cost for data delivery.Because short-range communication costs less than 3G/4G connectivity, the model encourages neighbor or WiFi forwarding.
  • Cost-based opportunistic transmission: The source-to-server opportunistic scheme can make global system cost arbitrarily close to the minimum under Lyapunov optimization conditions.The result depends on the system parameter multiplied by the average consumption term.
  • Relay selection: A relay model distinguishes integrated relays from courier nodes, forwarding critical packets through integrated relays and less-critical packets according to forwarding charge and threshold price.This pricing rule links packet criticality with the selected delivery mode.
  • Relay selection: Average consumption energy decreases by up to 40% versus AODV, but the approach does not define the threshold price for choosing courier forwarding over direct transmission.The missing threshold leaves a key forwarding decision unspecified.
  • Open direction: Combining adaptive compressed-sensing sparsity control with opportunistic transmission is proposed as an open direction for jointly reducing cost, energy use, and traffic.The surveyed text identifies this joint economic design as apparently understudied.

C. Relay Selection

Pricing and auction mechanisms select relay nodes for data forwarding while balancing energy use, link quality, route requirements, and network lifetime. Stackelberg-based relay pricing further coordinates relay competition and source decisions to improve utility and energy balance.

  • Motivation: Relay selection uses pricing as an incentive for rational forwarding nodes while sources seek routes meeting QoS, shortest-path, or energy objectives.Relay nodes may be selfish, so pricing supports data forwarding decisions.
  • First-price sealed-bid reverse auction: In the sealed-bid reverse auction, sources and neighboring relays submit asking prices based on relay and source information.Relay asks depend on hop count, link quality, and desired service price; the source ask averages its table information.
  • First-price sealed-bid reverse auction: The source buys from a neighbor whose asking price is below its own and selects the lowest asking price when multiple neighbors qualify.The simulation reports lower energy consumption than LEACH, whose single-hop strategy is identified as the reason for its weaker performance.
  • Dutch reverse auction: Dutch reverse auctions lower an initially high seller price iteratively until the buyer accepts, with the winner paid its asking price.The relay-selection model treats the source as buyer and relay nodes as sellers.
  • Stackelberg game: A Stackelberg relay-pricing game yields a unique equilibrium and achieves higher utilities and more balanced relay energy dissipation than simultaneous Nash strategies.Relays price with both competition and battery reserve capacity in mind.

1) Second-price sealed-bid auction:

Economic and pricing models address congestion, packet prioritization, packet dropping, and resource allocation in WSNs. The reviewed approaches improve forwarding decisions and delivery outcomes, but computational or communication overhead constrains deployment.

  • Second-price sealed-bid auction: Second-price sealed-bid auctions prioritize packets for congested transmission slots, while path prices can incorporate winning bids to select routes with smaller predicted utility loss.Auction information may support congestion management and dynamic path selection.
  • Second-price sealed-bid auction: Traveling auctions share local auction information among neighboring nodes to improve congestion decisions, but the added packet-header information increases network traffic.Headers include application ID, generation time, and delay.
  • Second-price sealed-bid auction: Price-based queue management prioritizes higher-price packets and reports better data delivery ratio than direct transmission, although priority computation can increase average delay.Prices combine priority dimension, priority class, and priority measure.
  • Value-based pricing: Value-based packet acceptance uses price and coverage fidelity to determine fair packet survival probabilities, producing higher throughput than FIFO especially during congestion.The accepting probability is used by the sink to select or drop packets.
  • Value-based pricing: Node price measures the total transmission attempts needed for successful delivery, allowing the sink to adjust reporting rates to reduce energy consumption while alleviating congestion.In dense networks, sending control information to every sensor is difficult when the sink is distant.
  • Combinatorial auction: Combinatorial auctions allocate bundles of complementary sensing resources through a market architecture, but winner determination is computationally complex.CABOB is exact but exponential in item count, whereas SGA has polynomial complexity in the number of bids and seeks feasible high-quality allocations.

2) Double auction:

Double auctions coordinate resource exchange by letting sensor tasks act as both buyers and sellers, while pricing models optimize rates, power, utility, and network lifetime under changing conditions.

  • Double auction: Double-sided auctions let multifunction sensor tasks exchange resources by submitting asks and bids to form supply and demand curves.The auctioneer uses these bids and asks to determine resource transactions and clearing prices.
  • Double auction: Non-cooperative game models can determine a Nash equilibrium when agents know others’ bidding information, but private strategies limit this assumption in real markets.The equilibrium gives no agent an advantage from unilateral payment-strategy deviations.
  • Double auction: Demand-and-supply rate allocation achieved higher peak signal-to-noise ratio than constant-rate allocation, which lacks knowledge of future demand.
  • Utility function: NUM-based rate allocation immediately adapts to each new system state, while step-size selection trades faster convergence against oscillations and stability.Newton’s method accelerates convergence, whereas larger step sizes can increase oscillations.
  • Utility function: Joint power-and-rate allocation uses Lagrangian prices and achieves lower link power with better data rates than utility-only MaxUtility.A related weighted utility model increases total utility while decreasing network lifetime as the weight rises from 0 to 1.

1) First-price sealed-bid auction:

First-price sealed-bid and related auction mechanisms dynamically allocate sensing tasks among sensors, users, and applications while addressing energy, efficiency, overhead, and truthful participation.

  • First-price sealed-bid auction: The Rust-In-Time market uses a first-price sealed-bid auction to negotiate task allocation among buyers, sensor sellers, a market entity, and resource agents.Sensors evaluate offered tasks using their current capabilities, enabling adaptation to network changes.
  • First-price sealed-bid auction: Participatory-sensing auctions account for seller distances and use maximum independent set approximation to prevent connected buyers from receiving tasks concurrently.The resulting allocation achieves the best-known bounded-degree approximation efficiency and strategy-proofness.
  • Reverse auction: Reverse auctions balance sensor energy and reduce consumption relative to static task allocation by allowing high-cost bidders to leave competition and sleep early.
  • Reverse auction: In mobile crowdsensing, reverse auctions use requesters, users, and servers for task publication, participant selection, price evaluation, and payment mediation.Posted-price mechanisms can replace users’ difficult-to-elicit true cost bids by learning reservation-price curves with k price arms.
  • Combinatorial reverse auction: Combinatorial reverse auctions share common tasks and resources across applications to reduce deployment costs, improve resource efficiency, maximize network lifetime, and maintain QoS.A two-phase winner-determination protocol reduces candidate asks before selecting suboptimal subsets heuristically.

5) Option pricing:

Pricing models support task valuation, resource-demand control, and sensing coverage by adapting allocations to uncertainty, supply-demand variation, energy limits, and coverage gaps.

  • Option pricing: Real option pricing values dynamic task-scheduling choices while accounting for time value and selection risk, with Black-Scholes used for strike-price determination.
  • Price discrimination: Rate-adaptive pricing changes per-unit resource prices using prior prices and current allocation rates, adapting task QoS to user-defined interest levels.The cited simulations did not report results across scenarios with different interest-level distributions.
  • Sensing coverage: Market-based coverage solutions achieve results similar to optimal centralized methods while substantially reducing computation cost and addressing scalability concerns.Centralized Hungarian-method computation is O(n3), where n is the number of sensors.
  • Area coverage: In area coverage, sealed-bid allocation moves mobile sensors to coverage holes and reduced required sensors by 30% when 10% were mobile.Static sensors bid according to hole sizes, while mobile sensors compare accepted prices with the holes created by leaving their current positions.
  • Target coverage: NUM formulates target coverage as aggregate-utility maximization subject to sensing-range, energy-consumption, and overlap constraints.

C. Barrier Coverage

Pricing models are applied to barrier coverage, target tracking, and security, using auctions to allocate sensors, reduce movement or energy costs, and isolate malicious nodes.

  • C. Barrier Coverage: First-price sealed-bid auctions assign mobile sensors to 2D barrier grid points with shorter maximum movement than the classic Hungarian solution.The approach reduces barrier-construction energy consumption but does not balance sensor energy when bids consider distance alone.
  • C. Barrier Coverage: For 3D barriers, auction-based sensor assignment must account for holes through which intruders can pass despite a formed 2D sensor chain.
  • D. Target Tracking: Auction-based target tracking saves more than 65% energy compared with a traditional dynamic coalition protocol without requiring prior neighbor knowledge.AASA further selects sensors using residual energy and distance to the predicted target location.
  • D. Target Tracking: Combinatorial auctions improve resource efficiency for multi-target tracking by bundling nearby targets into combined tasks under sensor resource constraints.
  • A. DoS Attack Prevention: Pricing-based secure routing selects paths that exclude malicious packet-dropping sensors, while reputation information can cause poorly behaving sensors to be ignored.For 100 sensors, total dropped packets remained below half the CONFIDANT protocol’s level in the reported example.

B. Privacy Concerns

Pricing models address privacy, incentive, and security concerns in participatory and crowdsensing applications. The reviewed approaches use auctions and cryptographic mechanisms to motivate truthful participation while protecting sensitive information.

  • Privacy incentives: Payments can motivate users to contribute sensing data when privacy concerns reduce their willingness to participate.A survey reported privacy importance averaging 5.82/7, while respondents preferred free provision, monetary rewards, or free data access in different proportions.
  • Privacy incentives: A sealed-bid second-price reverse auction selects participants with the lowest privacy valuation and pays winners the critical losing ask.The mechanism is intended to encourage sellers to report their true privacy values.
  • Privacy-preserving mechanisms: Privacy-preserving auction schemes conceal locations, identities, or asks using obscure locations, pseudonyms, encryption, signatures, and secure winner determination.Multi-attribute schemes can preserve anonymity, ask privacy, and public verifiability without opening asks during winner determination.
  • Secure sensing: Price-based sensing mechanisms can stimulate users to submit actual reports, while auction solutions can achieve optimality, rationality, and truthfulness in target tracking.One target-tracking formulation solves a multiple-choice knapsack problem to maximize fusion-center utility.
  • Open security issues: Existing auction-based security approaches address some denial-of-service and privacy issues, but intrusion detection and coexistence with malicious sensors remain open problems.The survey identifies these as future security research needs.

D. Pricing Models for Evaluating the IoT Deployment

Pricing models evaluate IoT deployment and service-market decisions across mobile sensing, supply chains, integrated hardware, and M2M communication. The reviewed methods combine economic analysis with optimization, auctions, and utility-based allocation.

  • IoT deployment evaluation: Cost-benefit analysis evaluates mobile-sensor deployment against static sensors by comparing economic utility while preserving functions such as area coverage.Mobile nodes are assessed according to their ability to replace multiple static nodes while maintaining network functionality.
  • IoT deployment evaluation: Supply-chain studies use cost-benefit models to analyze how deploying barcode, RFID, and sensor devices affects stakeholder profits.These devices monitor and improve goods transportation, linking deployment choices to supply-chain outcomes.
  • IoT deployment evaluation: Integrated WSN-RFID architectures distribute sensing and relaying tasks across hierarchical components but introduce additional hardware design and deployment costs.Cost analysis examines factors that can reduce expenses in these integrated networks.
  • M2M resource allocation: M2M pricing models also cover resource reservation, auction-based allocation, dynamic smart data pricing, and service-context communication among machine groups.These approaches treat devices or services as market participants seeking radio resources or reduced communication charges.
  • M2M resource allocation: Packet-attempt estimation before auction-based allocation raises transmission success probability to up to 99%, a 2% improvement over the earlier auction approach.Maximum-likelihood estimation predicts how many applications are attempting transmission so base stations can assign time slots more efficiently.
  • M2M resource allocation: A dynamic M2M rate-allocation model minimizes a concave utility function balancing service delay and price under RAN-capacity constraints.The algorithms converge to a unique rate and price solution, while simulations report lower delay-and-price utility than load balancing; Newton’s method may converge slowly.

X. SUMMARY, OPEN ISSUES AND FUTURE RESEARCH DIRECTIONS

The survey positions economic and pricing models as tools for both IoT business and system design, synthesizing applications across sensing, communication, security, and service markets. It identifies unresolved challenges involving mobility, bid security, payment privacy, complex markets, collusion, and computational overhead.

  • Summary: Economic and pricing models help network managers select schemes for objectives such as quality-of-service improvement and payment minimization.The survey organizes approaches by IoT issues and analyzes alternative economic mechanisms for system design and implementation.
  • Future research directions: Combining auctions with non-auction mechanisms, such as reverse Vickrey auctions and posted prices, may provide more flexible sensing schemes.The survey presents hybridization as a future direction rather than a demonstrated result.
  • Future research directions: Stationary-user assumptions in reverse-auction data collection require mobility tracking when selected phone users leave the area of interest.Lightweight triangulation can estimate future locations and include them among winner-selection attributes.
  • Future research directions: Future auction research must protect bid secrecy before opening and secure payment information, including identities and payments, from malicious users.Suggested directions include lightweight security protocols and bank-supported electronic cash schemes.
  • Future research directions: IoT resource markets should extend beyond traditional buyer-seller-auctioneer structures to support direct trades, inter-auctioneer transactions, buybacks, forfeiture, and lending.The survey identifies these diversified markets as future work.
  • Future research directions: Collusion among service providers can make Nash-equilibrium pricing inefficient, motivating analysis of collusion and the price of anarchy.Contract theory is also suggested to reduce relay-selection iterations caused by Dutch-auction approaches under asymmetric information.
Loading 1608.03475v1…