DYNA

2.1. IEEE 802.15.4

IEEE 802.15.4 [6,11,12] is a standard that deals with the intersection among low complexity, low energy consumption, low data rates, and low cost devices in WPAN. This standard specifies the physical (PHY) and medium access control (MAC) layers of the communication model (see Fig. 1). The PHY defines mechanisms for the: 1) activation and deactivation of radio transceivers, 2) channel energy level detection, 3) received packets link quality indication, 4) execution of Clear Channel Assessment (CCA) of the CSMA/CA protocol, and 5) transmission/reception of packets.

Figure 1: IEEE802.15.4 layers stack

Source: The authors

The MAC layer is in charge of channel access, maintaining a reliable link between devices, synchronizing, associating and disassociating devices, and employing a protocol for medium access control. There exist two mechanisms for medium access: beacon-enabled (slotted csma/ca) and non beacon-enabled (unslotted csma/ca).

In non beacon-enabled medium access, nodes keep listening to the channel and perform transmissions once they find it to be idle. Nodes can detect a collision when they do not receive an acknowledgement packet (ACK) from the receptor of the transmission.

In beacon-enabled medium access, a central device known as network coordinator periodically sends a beacon message used to sync up the network and to assign the transmission time slots for every node. Nodes contend for these time slots in the Contention Access Period (CAP) and enter sleep mode for energy saving purposes in the inactive part of the super-frame (see Fig. 2).

Figure 2: IEEE802.15.4 Slotted mode super-frame

Source: The authors

2.2. Cross-layer design

Cross-layer design basically performs a violation of the layered hierarchy by allowing protocols from non-adjacent layers to exchange information and relevant parameters obtained at run-time. The cross-layer approach proposes various alternatives to the traditional layered model: The creation of new interfaces to manage the connection and the flow of information among layers (see Fig. 3.a); the merging of adjacent layers, where a new "super-layer" is created by merging services provided by two or more layers (see Fig. 3.b); vertical calibration across layers, setting a priori information obtained from them or updating it based on parameters at runtime (see Fig. 3.c); new abstraction,

Figure 3: Cross-layer proposals and interactions.

Source: Srivastava and Motani, 2005.

which defines a non-layered architecture to have functions similar to those of the traditional protocol stack (see Fig. 3.d). Implementations of the cross-layer approaches have also been proposed based on direct communication between layers-(with exchanges of information between layers performed at runtime)-, a shared database across layers-(with a new layer-like scheme that performs the storage and retrieval of layer-related information), and a completely new abstraction (innovative implementations aimed at replacing the traditional stack like scheme) [8,9,13,14] (See Fig. 3.e).

2.3. Cross-layer protocol

The novel cross-layer technique for WBANs presented by Nabi et. al [9] is built upon 3 premises: 1) avoiding dependence on network topologies or network layer schemes for packet forwarding; 2) using a TDMA (Time Division Multiple Access) scheme as medium access strategy to avoid common energy wasting issues present in contention based schemes; and 3) adapting transmission power, since low power levels do not always ensure energy efficiency.

For routing duties, nodes in the network are supposed to have enough space in memory to store data items from neighbor nodes so each node broadcasts its own data item and at least a data item from any other node; this strategy is called “gossiping”. The power adaption mechanism makes the calculation of two metrics, one metric is concerned with outlink quality ( Q i ) and the other one prevents cluster formation when outlink quality improves among neighbors; hence it maintains a link to the sink node for every node in the network ( 𝐶 𝑖 ). This strategy cooperates with mobility given that isolated/disconnected nodes increase their transmission power to reach neighbors and reconnect to the network, because in isolation both metrics decrease consequently.

The authors claim that the aforementioned scheme ensures efficiency in energy consumption as it prevents retransmissions. Besides, data routing does not depend on the topology because the nodes broadcast information without concerning for the relative position of the intended receiver, and simultaneously, the latency is reduced because all nodes can reach the sink using the power adaptation mechanism. They also consider that the strategy needs to optimize listening on active time slots to reduce the consumption caused by idle listening.

Fig. 4 illustrates how power adaptation mechanism of the protocol works.

Figure 4: Cross-layer protocol power adaptation mechanism

Source: The authors

2.4. Energy consumption and QoS in WBAN

The tasks performed by nodes in a WBAN, and in general, in any WSN, including sensing, data processing, and communications incur in energy consumption. The wireless communication process constitutes the major power consuming task [15]. As nodes implanted on or deployed around the body have a very small form factor pursuing user’s comfort, batteries to power these devices are kept small, and hence energy storage capacity becomes a relevant issue. Most applications of WBANs restrict nodes to avoid requiring replacement of the battery over long periods of time; for example, a pacemaker or a glucose monitor is supposed to last for at least 5 years. Thus, energy-aware strategies have to be devised to keep nodes working properly over the required period of time. The network lifetime is defined as the time between the moment the entire WBAN is turned ON and the moment that the first node runs out of power, so energy saving strategies should support uniform energy drain, focusing on balancing power consumption among all the nodes comprising the network [16].

Another important issue is the constraint related to RF power, namely the consequences of the body's reaction to electromagnetic radiation, and the localized specific absorption rate in the body. To show the magnitude of the situation, sensors in a WBAN are required to broadcast data at low power due to heating of the radio chips, transferring heat to the tissues that form the muscles and causing harm to the WBAN user (burnings or even malignant cell degeneration) [17].

Besides the aforementioned issues, several causes of energy waste have been identified in WSNs, including collision, overhearing, idle listening, over-emitting, control packet overhead, and traffic fluctuations. Collisions occur when two or more nodes try to access the communication channel simultaneously. Idle listening is when a node listens to a channel lacking traffic. Overhearing occurs when nodes receive packets that are not intended for them. Control packet overhead happens when a node adds too many control headers to the payload. Finally, over-emitting means that many retransmissions are required for a packet to reach its destination [18].

QoS’s guarantees in WBAN need special attention given the context of application of this type of network. For instance, in medical environments, sportsmen and soldiers vital signs monitoring. Although WBANs as a subset of WSNs face the same technical challenges of most wireless networks, it is important to notice that the critical levels regarding issues like network connectivity, security, delay, energy consumption, packet delivery, traffic prioritization, and bandwidth allocation are more demanding in WBANs [19].

In this research, we pretend to compare two protocols for WBAN, a cross-layer scheme for WBAN and IEEE 802.15.4 standard. We focused on the tradeoff between energy efficiency and QoS accomplishment in both protocols. The following three QoS metrics are in the scope of this work: End to End Delay, Packet Delivery Ratio, and Goodput as they span overall major requirements in WBAN transmissions.

3.1. Experimental design

The experimental design is a necessary stage to achieve the goal of a fair comparison between the protocols for WBAN. The main focus of our work is to assess the energy consumption in both protocols and analyze which of them is more suitable for WBAN, given the constraints introduced by the management of the battery. This analysis, however, is not enough since there is a tradeoff between energy consumption and acceptable QoS metrics in the context of WBAN. Hence, the evaluation was aimed to reveal which communication scheme had the lowest power consumption and the best QoS accomplishment.

The experiment was devised as a model of fixed effects, in which we chose all the treatments. The results and conclusions are assumed to be tightly tied to the levels of each factor involved in the experiment. To conclude about the results of the experiment, we proposed a set of hypotheses. A major requirement of the experiment was to run a fixed number of replicas of each treatment to include the effects of variability induced by nuisance factors and run these replicas randomly to make the environment of treatment execution as uniform as possible as in completely randomized experimental designs [20].

3.1.1. Experimental unit

The experimental unit is a wireless body area network comprised by the following components, as shown in Fig. 5: 1) a sink node and a few sensor nodes installed on a patient’s body. All of the nodes have specifications based on the TI-CC2420 radio [21], whose model is available in OMNeT++; 2) a wireless channel prone to path induced attenuation and signal shadowing (lognormal); 3) a mobility model used to simulate the patient’s movement (we used MoBAN [22]).

Figure 5: Experimental unit, factors, constants, and response variables of the experiment.

Source: The authors

3.1.2 Factors

The factors listed below were considered to have significant influence on the response variables as they represent the main characteristics of the WBAN itself and also by taking an experimental reference framework for wireless networks from [5,23,24].

• Communication scheme (protocol): This factor has two qualitative levels, one corresponds to the cross-layer protocol and the other one to protocols of the IEEE802.15.4 standard.

• Number of nodes: We considered this factor to have direct incidence on QoS metrics and power consumption, given that network connectivity could be impacted by the amount of neighbors surrounding individual nodes and the distance between them. Two levels were defined for this factor, networks of 5 and 12 nodes.

• Traffic in the network: As we wanted to observe the power consumption in the WBAN, we considered that the amount of traffic could have a direct impact on battery management in wireless transmissions. We defined two levels for this factor, as follows: one level corresponds to heavy traffic (nodes transmitting every 30 seconds) and low traffic (transmissions are performed every 5 minutes or 300 seconds).

As shown in Table 1, the treatments of the experiment are all combinations of all levels of the factors, and every treatment was replied 10 times in the experiment. As referred before, the replication order was randomized in order to keep independence and uniformity between treatments.

Table 1: Experiment treatments

Source: The authors

3.2. Simulation setup

The simulation tool used in our work was OMNeT++, a modular discrete event simulator that is suitable for telecommunication networks and protocol simulations. The networks can be modeled using NED (Network Description) language and the behavior of protocols and modules is coded using C++. OMNeT++ has a special framework for wireless network simulation called MiXiM [25], [26], which provides useful built-in modules and models, like battery, channels, packets, messages, mobility, PHY, MAC, Network and Application layers, and several examples for wireless networks.

4.1. Cross-layer and IEEE802.15.4 unslotted

Fig. 6 shows the results, in terms of means, for the response variables comparison between the cross-layer protocol and the unslotted mode of IEEE802.15.4.

Figure 6: Average performance of the QoS metrics in cross-layer VS. IEEE802.15.4 unslotted

Source: The authors

For average end to end delay and average packet loss rate response variables the null hypothesis was rejected, the graphics show the difference between the means obtained with both levels of Protocol factor. Regarding the End to End Delay metric, it is shown that IEEE 802.15.4 exhibits a slightly better performance. The difference in the means was close to 2.5ms, which is not significant in WBAN environments.

With regard to the Packet Loss Rate, the IEEE 802.15.4 unslotted outperforms the cross-layer protocol. This can be explained by the amount of lost packets that the later protocol exhibits when all nodes are adapting power in the early stages of the network.

Regarding Goodput, the results match reality since the network using the cross-layer scheme broadcasts larger packets due to the “gossiping” strategy, with packet loss and end to end delay not far from those exhibited by IEEE 802.15.4 unslotted.

Regarding Energy Consumption, the results are totally conclusive, nodes using cross-layer protocol saved almost 90% of the battery in the entire simulation run, while when using IEEE802.15.4 protocols only a 10% of the battery remained untouched.

4.2. Cross-layer and IEEE802.15.4 slotted

Means plots in Fig. 7 show the results for the response variables comparison between the cross-layer protocol and the slotted mode of IEEE 802.15.4.

In this case, the only null hypothesis rejected is the one regarding the remaining battery load in the network. The difference between the response variables means for the rest of hypotheses is quite appreciable. For the End to End Delay, Packet Loss Ratio and Goodput metrics, a large difference is noticeable between the protocols. The best performance in those metrics is exhibited by the cross-layer protocol.

Figure 7: Average performance of the QoS metrics in cross-layer VS. IEEE802.15.4 slotted

Source: The authors

However, regarding the Battery Consumption, a difference close to 10% in favor of IEEE 802.15.4 slotted can be observed. This result can be explained by longer inactive periods in IEEE 802.15.4 slotted and less slots in listening mode in IEEE 802.15.4 unslotted when compared with the cross-layer ones.

4.3. Transmission power

Another important issue that we considered in our experiments was the transmission power level. As mentioned in section 2, WBAN nodes should not perform transmissions at high power levels, not only aiming to reduce energy consumption, but also trying to reduce the electromagnetic absorption rate in the body, which could result in health problems related to WBAN’s use.

Figs. 8-11 show the transmission power levels for every node in 4 randomly chosen treatments of the experiments, Figs. 8, 9 correspond to WBANs formed up by 5 nodes and transmissions every 30 seconds, and 5 minutes respectively, while Figs. 10, 11 correspond to cases with 12 nodes.

Figure 8: Performance of WBAN nodes regarding transmission power levels (network of 5 nodes - sending packets every 30 seconds)

Source: The authors

Figure 9: Performance of WBAN nodes regarding transmission power levels (network of 5 nodes - sending packets every 5 minutes)

Source: The authors

Figure 10: Performance of WBAN nodes regarding Transmission Power levels (network of 12 nodes - sending packets every 30 seconds)

Source: The authors

Figure 11: Performance of WBAN nodes regarding Transmission Power levels (network of 12 nodes - sending packets every 5 minutes)

Source: The authors

It is evident that in networks using the cross-layer approach, nodes performed transmissions at lower power levels help to reduce electromagnetic absorption. It can be observed that in WBANs with a greater density of nodes, the power level tends to reduce considerably among all nodes, even performing transmissions at the lowest power levels.

Such a result shows an extra benefit resulted from the use of the cross-layer protocol, as it prevents collateral damage and discomfort to WBAN’s users.