Energy consumption resulting in global CO2 emission and battery waste caused by data communication and networking devices is increasing exponentially. Information and communication technology (ICT) is responsible for about two percent of the global CO2 emissions. However, ICT includes Internet of things (IoT) technologies and applications that have a direct effect on lowering CO2 emissions by increasing energy efficiency, reducing power consumption, and achieving efficient waste recycling. IoT has, therefore, an interesting dual role in CO2 emission [CarbonRoom] [Vermesan 2011] since its developments show that we will have 16 billion connected devices by the year 2020 (i.e. average out to six devices per person on earth and to many more per person in digital societies) [Vermesan 2011]. A recent report by the Carbon War Room [CarbonRoom] estimates that the incorporation of machine-to machine communication in the energy, transportation, and agriculture sectors could reduce global greenhouse gas emissions by 9.1 gigatons of CO2 equivalent annually. Nowadays there is a need to develop applications that are environmentally friendly. In this context, the GRECO project (GREen wireless Communicating Objects) proposed to study the design of autonomous communicating objects. This means that, within a given time period, the power consumption is lower than or equivalent to the energy the object can harvest from its environment. The approach developed in GRECO aims at reaching a global power optimization for a communicating object.
Wireless Sensor Networks
Wireless sensor networks (WSN) are dedicated to short range wireless communications with low energy consumption. They are designed to handle a small amount of transmitted data generally corresponding to battery-operated sensors measurements (e.g. temperature, pressure, security cameras, controllers for water sprinklers, etc.). They have limited memory and computational capacity. Their use has become widespread in industrial environment, in home automation systems and in military systems [Garcia 2007]. Standardized protocols and proprietary protocols have been proposed. Based on the comparison given in Table 1, the ZigBee protocol has proved to be simpler than the Bluetooth, UWB, and Wi-Fi, which makes it very suitable for sensor networking applications. Other proprietary protocols have been used in sensor networks. However, for the sake of the interoperability between devices, a standardized protocol is preferable.
The corresponding network topology. Almost all protocols use the 2.4 GHz ISM band. This common characteristic can be considered as a first level of standardization. Therefore, it should also be mentioned that even a proprietary protocol has to be in compliance with a number of rules and meet some compatibility requirements such as the national authorities’ regulation for radio transmission.
Many constraints are considered when designing and deploying sensor networks. The main constraints are related to the energy consumption, the production cost, the hardware limitation, the operating environment, the network topology, the range, the transmission medium and the throughput. This is the reason that makes compromises in term of data rate essential. Depending on the application, additional constraints are added such as mobility management, energy management, etc. Some of these constraints (low consumption, high throughput and large range) are so contradictory that it will never be a unique standard since different solutions can be considered. This gives space to researchers to propose different approaches. Low consumption, high throughput and large range are important factors and are usually used as guidelines to develop algorithms and protocols used in sensor networks. They are also considered as metrics for comparing performance between different solutions in this field.
Challenges in WSN
Communication between sensor nodes faces more challenges than communication in other wireless networks. The constraints imposed by sensor architecture makes it hard to ensure a high quality of service (QoS). Besides, the low signal power used to transmit data makes the signal very vulnerable to channel disturbance. The main challenges to which sensor networks are confronted are mainly the low energy budget, the channel conditions, the collisions that may occur during the packet transmission and the mobility which has to be ensured in many WSN applications. All these features are detailed below.
Each single execution within electronic devices needs energy. This is why energy consumption optimization is a matter raised at every level from the sensor architecture design phase to the phase of WSN protocol applications’ conception. Besides, WSN nodes are supposed to work autonomously for a long period of time, a low energy budget will impose in certain cases adjustments in the system functionalities in order to reduce the energy consumption [Shah 2002]. The energy budget is, then, both an evaluation metric that gives information about some algorithms performance and a decision metric used by protocol algorithms.
The transmission unit consumes the biggest part of the energy [Raghunathan 2002]. Indeed, datasheets of commercial sensor nodes show that the energy cost of receiving or transmitting a single bit of information is approximately the same as that required by the processing unit for executing a thousand operations [Crossbow] [Tmote]. To overcome this shortcoming, the sleep and the idle modes are used in sensor nodes. In the sleep mode, significant parts of the transceiver are switched off. The node is not able to immediately receive data and needs a recovery time to leave the sleep state (energy consumed during the startup can be significant). In the idle mode, the node is ready to receive, but it is not doing so. Some functions in the hardware can be switched off, thus, reducing the energy consumption. The use of sleep and idle modes raises a new problem which is the synchronization of network nodes. Devices need to know when to switch between the different states: receive, transmit, idle and sleep modes. Moreover, a clock drift may occur [Ganeriwal 2005]. In this case, control packets have to be transmitted to resynchronize the network, which increases the energy consumption. The major goal when designing WSN applications is to ensure the optimal throughput with a low energy budget according to the targeted application requirements [Chandrakasan 1999].
Signal distortion during the packet transmission can be caused by predictable and quantifiable phenomena (at least when the transmission environment is well known) and by unpredictable events. The main predictable phenomena are the propagation which consists in the attenuation of the transmitted signals with the distance (path loss), the blocking of signals caused by large obstacles (shadowing), and the reception of multiple copies of the same transmitted signal (multipath fading). These variations can be roughly divided into two types [Tse 2005]:
• Large-scale fading, due to path loss of signal as a function of distance and shadowing by large objects such as buildings and hills. This occurs as the mobile moves through a distance of the order of the cell size (cellular system), and is typically frequency independent.
• Small-scale fading, due to the constructive and the destructive interference of the multiple signal paths (multipath fading) between the transmitter and receiver. This occurs at the spatial scale of the order of the carrier wavelength, and it is frequency dependent.
Large-scale fading is more relevant to issues such as cell-site planning. Small-scale multipath fading is more relevant to the design of reliable and efficient communication systems. Unpredictable events happen randomly and are completely unknown by the device. The corresponding effects are the thermal noise and interferences. The thermal noise is introduced by the receiver electronics and is usually modeled as Additive White Gaussian Noise (AWGN). If the medium is not shared with any other RF sources, the signal propagation of simulated transmitters and AWGN can describe the entire channel. However, when several nodes transmit at the same time on the channel, interferences may happen and have to be taken into account in the channel modeling.
In WSN, the transmission power is relatively low, which makes the signal very sensitive to noise. In addition, since the antennas used by the nodes are very close to the ground, the loss of the signal transmitted between them can be very high.
In [Tse 2005], it was highlighted that interference I from neighboring cell is random due to two reasons. One of them is small-scale fading and the other is the physical location of the user in the other cell that is reusing the same channel. The mean of I represents the average interference caused, averaged over all locations from which it could originate and the channel variations. However, due to the fact that the interfering user can be at a wide range of locations, the variance of I is quite high. Therefore, it was noticed in [Tse 2005] that the signal to interference plus noise ratio (SINR) is a random parameter leading to an undesirably poor performance. There is an appreciably high probability of unreliable transmission of even a small and fixed data rate in the frame.
Table of contents :
Chapter 1. Introduction
1.1. General context
1.2. Short range protocols
1.1. Wireless Sensor Networks
1.2. Challenges in WSN
1.2.1. Energy consumption
1.2.2. Channel conditions
1.2.3. Medium access
1.3. Contributions and manuscript organization
Chapter 2. IEEE802.15.4/ZigBee overview
2.2. Sensor node architecture
2.2.1. The sensing unit
2.2.2. The processing unit
2.2.3. The transmission unit
2.2.4. The power supply unit
2.3. IEEE 802.15.4 overview
2.3.1. IEEE 802.15.4 versions
2.3.2. IEEE 802.15.4 nodes
2.3.3. IEEE 802.15.4 topology
2.3.4. Physical layer
2.3.5. Mac sublayer
2.3.6. Logic Link Control (LLC) sublayer
2.4. ZigBee protocol [ZigBee]
2.4.2. Addressing mode
2.4.3. ZigBee routing protocols
2.5. Energy consumption in IEEE 802.15.4 WSNs
2.5.1. PHY Layer
2.5.2. MAC Sublayer
Chapter 3. State of the art
3.2. Proposed approaches for reducing energy consumption
3.2.1. PHY Layer
3.2.2. MAC sublayer
3.3.1. Ad hoc networks
3.3.2. Mobility in cellular networks: different solutions
3.3.3. Mobility in IEEE 802.15.4
3.3.4. Mobility models
3.4. Link Quality Estimation
3.4.1. Considered parameters
3.4.2. Overview of link quality estimators
3.5. Rate adaptation in IEEE 802.15.4
3.5.1. Available rates in IEEE 802.15.4
3.5.2. Overview of proposed rate adaptation algorithms for IEEE 802.15.4
Chapter 4. An Enhanced Mobility Management Approach for IEEE 802.15.4 protocol
4.2. Proposed network architecture
4.2.1. Network topology
4.2.2. Addressing and routing
4.3. Simulation tools and general simulation setup
4.4. Energy consumption in the standard procedure
4.5. Handover procedure
4.6. Efficiency of using LQI in mobility management
4.6.1. Network architecture and initialization
4.6.2. Selection of the new coordinator
4.6.3. Simulation use cases
4.6.4. Gain in energy and delay for both scenarios
4.7. An LQIthreshold formula for mobility management
4.7.1. LQIthreshold formula
4.7.2. Impact of β parameter
4.8. Same-road speculative algorithm
4.8.1. Simulation setup
4.8.2. Evaluation of the proposed approach
4.9. Probabilistic speculative algorithm
4.9.1. Description of the Probabilistic Algorithm
4.9.2. Gains in energy and delay of the probabilistic algorithm
Chapter 5. Rate adaptation algorithm
5.2. A Mobility-aware Rate Adaptation Algorithm
5.2.1. Rate selection algorithm
5.2.2. Evaluation of the Approach
5.3. An adaptive rate adaptation algorithm
5.3.1. Link Quality Estimation Metrics based on chip error rate
5.3.2. Rate Selection Algorithm
5.3.3. Scenario description and simulation setup
5.3.4. PDR evaluation and mobility impact
5.3.5. Energy efficiency
Chapter 6. Conclusion