This paper addresses the low power mechanisms provided by the ZigBee and the 6LoWPAN Protocol, providing comparative assessments based on the results obtained by di erent researchers and available in specialized literature, running through experimental measurements on digital test banks. For a performance comparison, the parameters of each protocol have been adjusted so that it is able to function properly in low power mode and make the measurement scenarios equivalent in terms of tra c and energy e ciency. The comparison focuses on the impact of the mechanisms of low power in the performance of the network. Experimental evaluations mentioned, show strengths and weaknesses of both protocols when working in a low power mode.
The gravitational nucleus formed by the concepts of Web 2.0 and the Internet of
things (IoT) is moving towards Web applications and concepts of services looking
for greater integration and greater accessibility, in the light of new paradigms of
"at any time" and "at anywhere". The future of the Internet is based on these
"bricks" willing to form a dynamic entity, producing new means of interaction
with the services, other users and the environment [
The present study aims to compare the performance of the modes of low
consume operation of IEEE 802.15.4 / ZigBee and 6LoWPAN, for a wireless
network of industrial sensors - WSN. Therefore the rst part describes the two
protocols with an approach to low power mechanisms that are implemented and
their capabilities. As it follows we cited the results of actual comparisons of both
protocols in a digital test bank, focusing on the impact of the mechanisms of
low power on the performance of the network. Since these mechanisms of
lowwork power put some nodes at rest in order to decrease their cycle, the protocol
comparison is made by making them work on the same platform HWSW,
under the same workload and using the same cycle. To accomplish this, we need
a preliminary adjustment of the operating parameters of the two protocols. Such
tuning is not simple, just as the mode of low power consumption implemented
by the two protocols, and therefore, the relevant parameters are signi cantly
di erent [
WSN Internet interconnection has been widely investigated in recent years. At
the beginning of the WSN era, the researchers focused on the development of
dedicated systems, then very specialized, but non-standardized protocols were
designed. Although these systems were generally e cient in speci c application
scenarios, they lacked exibility: the development of new applications required
much time and it was a very cumbersome job. As a remedy to this, the standard
6LoWPAN is proposed as a viable method to carry IPv6 WSNs [
The 6LoWPAN Protocol is attracting the attention of the market and the
researchers, thanks to the new possibilities with the integration of WSN and
Internet technologies. In [
are snuobdmesit[7te].d by a coordinator in order to synchronize the associated nodes (IEEE, 2006). The superplots which are shown in Figure 1 are divided into two parts: the active part and Tthheeisnuapcetirvpeloptsarwt.hTichheaarectsihvoewpnarint iFsidgiuvried1edarientdoiv1i6decdoinnnteocttworospoafrttsh:ethsaeme length, aancdtivbeyptahret lainmditthoefinthaecstievenopadrets.,Tthe acchtaivnenpealrctains dbievidaecdceinssteod16tocotnrannescmtoirts ionfformation. Tthhiessiasmdeivleidnegdthi,nantudrnbyintthoe tlwimoitsuofbpthaerstes:ntohdeeCs,otnhteacinhamnennetl AcacncbesesaPcceersisoedd (tCo AP, for its atcrraonnsymmit iinnfoErmngaltiisohn).,Tinhiswishidcivhidthede imnteudrina ianctocetswso issubrepgaurtlsa:ttehde bCyontthaeinCmSeMntACA, and thAeccPersosoPfeprieordio(dCAofPc,ofonrtaiitnsmacernotny(PmPCin, Efonrgliitssh)a,cirnonwyhmichinthEenmgleidshia),aicncewsshiisch you can regulated by the CSMACA, and the Proof period of containment (PPC, for its access the channel without dispute. In the inactive part, on the other hand, there is no acronym in English), in which you can access the channel without dispute. In the inactive part, on the other hand, there is no communication and therefore the nodes can turn their radios out and enter (Suspension) energy saving mode. The timing of the superplot is governed by two parameters: { The order between the Bacons (BO) or guidelines, which de ne the interval between two guides, which consequently de ne the duration of the superplot.
This interval is called interval between guides (BI) and is de ned as BI= 2BO { The order of the superplots (SO), which de nes the active period of the same duration. In accordance with the IEEE 802.15.4 standard, must be de ned in the following order: 0 SO BO 14.
As the nodes sleep in the period of inactivity of the superplot, the duty cycle (DC) of this only depends on the structure of the superplot, and it can be speci ed as:
DC = SBDI = 2(IO) (1) Where IO = SO BO is called the order of inactivity.
When using IEEE 802.15.4 devices in their low-power protocols with the
Guide element enabled, two data of di erent transfer protocols are used to
achieve communication between a coordinator and its devices. In the case that
a packet must be sent from a device to which the one that acts as Coordinator,
it is used a direct-drive mechanism. In this Protocol, the device waits for the
plot Guide, sent by the Coordinator, to synchronize with the superplot. Then,
it sends data from one of the slots of the superplot. To run the reception of
the package, the Coordinator sends an acknowledgement of receipt to complete
the transmission. This last point, however, is optional. Two di erent forwarding
protocols are de ned by the ZigBee speci cation: a protocol for routing based
on the Ad hoc on - demand distance of the Vector (AODV, for its acronym in
English), and a routing protocol tree which forwards the packets to the following
hierarchy of coordinators. Topologies of star and group of trees can take
advantage of the feature of low power consumption. However, the star topology can
only be used for small networks, while tree cluster topology is suitable for larger
sensor networks. For this reason, this document focuses on the topologies of the
trees with tree routing cluster [
The 6LoWPAN [
Taking into account that the maximum size of plot in the MAC layer is 102 bytes, or even smaller when the link layer security is enabled, just a few bytes would be available for data transmissions. On the other hand, IPv6 minimum maximum transfer unit is 1280 bytes (RFC 2460), which is much larger than the maximum for the IEEE 802.15.4. Therefore, it needs a little fragmentation. Another of the problems raised by the support of IP over IEEE 802.15.4 MAC and physical are the scheme of addressing and the routing mechanism. In fact, IP and IEEE 802.15.4 have di erent protocols for address and routing that cannot handle nodes in sleep mode. Therefore, 6LoWPAN introduces an adaptation layer between the MAC and the network layer, which compresses the header to a few bytes, keeping the main features of IPv6. The guarantee of adaptation of 6LoWPAN layer, introduces support mechanisms to the fragmentation and to the reassembly, when compressing the headers, and carries out the assignment of addresses between IPv6 and IEEE 802.15.4. The details of these mechanisms, however, are outside the scope of this work.
The 6LoWPAN speci cation does not provide any speci c mechanism to achieve low-power communication, but it depends on the bottom layer approaches.
The simplest solution to achieve energy e ciency in 6LoWPAN networks could be the exploitation of the low-power capabilities of the IEEE 802.15.4 protocol with enabled Guide. However, while the MAC layer could theoretically work both in enabled mode as in not enabled, all 6LoWPAN current implementations only support the Protocol not enabled. The main reason for not applying the communication without a guide in 6LoWPAN is that IP protocols assume that the link is always active. This is for the sake of simplicity, so that IP protocols do not need to program transmissions of datagrams. Therefore, 6LoWPAN uses techniques that give the illusion that the upper layers in the receiver are always active, although actually it only happens part of the time. Most of 6LoWPAN implementations use low powers of listening to power (LPL), a technique that allows the nodes to access channel asynchronously in a fully distributed network.
The Operation behind LPL is as follows. A node periodically wakes, lights and so do the activity controls. If no activity is detected, the node turns o the receiver and returns to its state of rest, or otherwise, the rooms stay turned on until the packet is received or, in the case of a false positive, until a timer expires. A graphical representation of the LPL operations is shown in Figure 2. 4
Benchmarking is done in two di erent test scenarios, in which the nodes have the same con guration, but the sending node transmission period (and therefore workload) is di erent. The rst scenario emulates an application of detection of low power where the monitored phenomenon has slow dynamics, and where the sensor node sends packets of data with a period of 20s. The second scenario emulates a WSN application with a faster dynamics, where the shipping period is 400ms. It should be noted that the di erence in shipping period also changes for the 6LoWPAN`s working cycles. The topology of the network, as in the scenarios is the same as shown in Figure 3. All nodes in the IEEE 802.15.4 ZigBee network share the same superplot settings and, therefore, the same working cycle. Four di erent settings of superplot have been used for the net IEEE 802.15.4 ZigBee, with di erent working cycles and, therefore, di erent energy saving capabilities. The rst setting uses a BO of 3, which is a smaller guide, which o ers su cient security of transmission, with this BO we use an OS, which in turn is the smallest SO that provides su cient reliability of transmission. We get a cycle of 0.25. To further reduce the duty cycle, the superplot is set using a Guide order of 5. In this way, we have been able to work down to 0.06 cycles. The details of these con gurations are shown in Table 1.
The 6LoWPAN settings (BI and SI) have been selected to provide the emitter node with the same working cycle as the IEEE 802.15.4 / ZigBee network. It also depends on the shipping period, so we use di erent con gurations in the two scenarios. 4.1
Scenario 1 - Tsend = 20s
DC =
AI SI + AI +
Tsend
(1)
With SI (interval of inactivity) as unknown and the value speci ed for the other
parameters, according to [
The experiment consisted of sending 1300 packages and record the times of transmission and reception of each packet. The log les are then used to calculate three performance metrics:
The delay from end to end, calculated as the di erence between the date and time of an event of reception and the date and time of the event of corresponding transmission.
The time of update is the time interval between two consecutive events for the reception of packages.
The relationship of delivery, which is the percentage of received packets over the total number of sent packets successfully.
The cumulative distribution function (CDF) of the experiments in delays from end to end for all measurement activities are represented in Figure 3, while the average delay and standard deviation are shown in Table 4. Here it is possible to see that the delay from end to end of the 802.15.4 / ZigBee IEEE network is strongly in uenced by the beacon interval. In fact, the ZigBee con guration - 1 with a Guide order of 3 has less delay than in the ZigBee - 4 con guration that has the same working cycle but with a Guide order of 5. The delay from end to end of the 6LoWPAN con gurations are also very in uenced by the cycle of work regarding the network con gurations with 0.25 and 0.12 duty cycle, we can see that 6LoWPAN has a smaller average delay, but with a considerably larger standard deviation than IEEE 802.15.4 ZigBee.
In Figure 3 you can see that, taking into account the ZigBee - 1 and 6LoWPAN - 3 scenarios with a duty cycle of 0.25, although the average delay is smaller, in this last protocol is larger, the ZigBee delay - 1 is approximately of 200 MS, which is much smaller than the 6LoWPAN - 3. However, when considering the scenarios that have a cycle of 0,12 (ZigBee - 3 y 6LoWPAN-2), the 6LoWPAN Protocol provides a smaller delay than IEEE 802.15.4 / ZigBee most of the time. Finally, from Table 4, it is possible to observe that the decrease in the length of the AI in the 6LoWPAN - 1 con guration, has a serious negative impact on the performance of the network, since both, the delay and the increase in the standard deviation of some of them will mean an order of magnitude. As a result, the ZigBee - 2 con gurations with the same cycle of 0.06 clearly exceeds 6LoWPAN - 1.
In terms of the CDF of the times of update, in Figure 4, we see that IEEE 802.15.4 / ZigBee achieves smaller time intervals between consecutive packets receptions. Due to the fact that TelosB nodes have a slightly shorter symbol time than the expected by the standard, the actual shipping period is shorter than 20s, and in particular is about 19, 07s. Therefore, the smaller update times of the IEEE 802.15.4 / ZigBee con guration are closer to the 6LoWPAN. Once again, the 6LoWPAN - 1 con guration shows much better performance than the other con gurations. Finally, we see in the ZigBee { 1 con guration, that there are some sporadic cases where the update time is twice the theoretical value. A closer look at Figure 4 reveals that this also occurs for the other con guration of IEEE 802.15.4 / ZigBee, but with a lesser probability.
This result is due to packet loss, if a packet is lost, the update time is recorded in the next reception of packets, and therefore doubled the time of update. Regarding the relationship of distribution, in Figure 5, we see that the 802.15.4 / ZigBee IEEE network experiences some loss of packets, which increases when the order guide decreases. One possible explanation for this behavior is that this packet loss is due to some problems of synchronization due to the variability of real beacon intervals. In the gure we see also that 6LoWPAN with the default value of IA (51.2ms) has the best performance in terms of packets received (but the ZigBee with BO = 5 con guration is just marginally less). On the Opposite, the 6LoWPAN -1 with the smaller AI settings has better performance in all aspects. The smaller shipping period of this scenario makes the 6LoWPAN- 1 con gurations and 6LoWPAN - 2 to be unusable, because their sleep intervals are comparable to the shipment times, and this would lead to congestion in the network. Con guration 6LoWPAN - 3 is still usable, but the comparison with the other networks would be unfair, since the cycle of this con guration is 0.7 considering the 400ms of shipping in each period. It is possible to see that 0.25 or smaller working cycles are not reachable at all with this period of shipping and with the IA by default. On the other hand, it was seen that the decrease in the duration of active intervals causes a degradation of performance, for this reason, it was decided to compare the performance of the same con guration of 802.15.4 / ZigBee IEEE stage 1 against the 6LoWPAN settings that provides the smallest work cycle for the shipping, it is possible to see that the optimum is a SI of 92ms which provides a 0.64 duty cycle with the default value of AI 51.2ms. We refer to this network con guration such as 6LoWPAN - 4. The same performance data observed for stage - 1 were also calculated for the stage - 2.
In the CDF of end-to-end delay in packets, we see basically the same trends that were observed for the stage - 1 the delays of the IEEE 802.15.4 / ZigBee are almost identical to those shown in Figure 3. 5
Several authors have examined the performance of 6LoWPAN and ZigBee protocols independently. A project os IETF 15 assesses the ZigBee Protocol whereas several routing scenarios of real-life. Malisa Vucinic, Bernard Tourancheau and Andrzej Duda, made a comparison of the performance of the RPL and LOADng protocols proposed by the IETF for low-power and high-complexity based networks on the IPv6/6LoWPAN.
ZigBee is a network layer built on the IEEE MAC 802.15.4 standard, designed to provide a protocol based on the standards for the interoperability of the sensors networks. 6LoWPAN, the new buzz word, seems to be competing heavily with ZigBee.
In essence, 6LoWPAN products have entered the market as the ZigBee direct competition, since they can use the mentioned standard (802.15.4), and even better, they can work with other PHY and allows seamless integration with other IP-based systems. 6LoWPAN is an acronym of IPv6 via Low-Power Wireless Personal Area Networks; the name originated in the Working Group in the IETF.
{ 6LoWPAN vs. ZigBee: Wireless Protocol Interoperability ZigBee de nes communication between 802.15.4 nodes (Layer2 in the IP World), and then de nes new upper layers for all the way to the application. This means that ZigBee devices can interoperate with other ZigBee devices, assuming that they use the same pro le (similar to Bluetooth). Besides the 6LoWPAN Protocol provides interoperability with other wireless 802.15.4 devices as well as devices in any other IP network link (for example, Ethernet or Wi-Fi), with a simple bridge device. To build a bridge between non-ZigBee and ZigBee networks requires a more complex gateway at the application layer.
The key requirement for IPv6 on 802.15.4 is that the Maximum Transmission Unit (MTU), must be at least 1280 bytes packages (by RFC 2460). Since the IEEE 802.15.4 standard packet size is 127 bytes, an adaptation layer must be implemented to allow the transmission of IPv6 datagrams.
{ 6LoWPAN vs. ZigBee: Size of Stack / Overload of Packages When comparing ZigBee and 6LoWPAN over 802.15.4, the professional should be familiar with the package format and general costs, since this is directly related to the ease of scaling the network and the available space for payload data. Although there are alternative ways, typical con gurations are shown below.
The IP routing over 6LoWPAN links, does not necessarily require additional header information in the 6LoWPAN. Layer. This reduces the overload of packets and allows more room for the payload data. Moreover, the size of the typical code of a pile with all the functions is 90 KB for ZigBee and only 30 KB for 6LoWPAN.
Fctrl: Frame control bit elds; Dep: Destination endpoint; Clst: Cluster identi er; Prof: Pro le identi er; Sep: Source endpoint and APS: APT counter (for the prevention of duplicate sequence).
{ 6LoWPAN vs. ZigBee: Availability and Cost The major players in the industry of semiconductors, such as Texas Instruments, Freescale and Atmel, promote, and provide 802.15.4 tabs that can be used, either for ZigBee or 6LoWPAN. These same companies even o er free ZigBee stacks. The Support for 6LoWPAN stacks seems to be behind ZigBee. There is at least one available open source stack, and some companies such as Archrock and Sensinode licensed their own 6LoWPAN stacks. In summary, 6LoWPAN is quite attractive, since it is based on IP - the standard work of Internet Protocol. However, ZigBee seems to be more popular and has been adopted by the main actors in multiple industries.
This work was directed to an evaluation of the comparative performance of the IEEE 802.15.4 / ZigBee and 6LoWPAN protocols for industrial WSNs. The document o ers several contributions. First, a theoretical analysis of the low-power characteristics of the two protocols, which are used as the basis of a methodology to tune the con guration parameters of low-power of the two protocols (it means., the BO and SO for 802.15.4 / ZigBee, and the AI and SI for 6LoWPAN) in order to achieve a working cycle in a realistic test bank built with COTS Hardware and open source software. And seconds, the experimental comparative evaluation in a digital test bank, that allow to highlight the strengths and the weaknesses of both protocols when working in the mode of low consumption. From these assessments arise tips for the designer of the network, allowing him to select the technology that best suits the purpose of performance, to properly achieve a given duty cycle, or a maximum delay.
The 802.15.4 IEEE protocol allows a network designer to carry out energy planning by adjusting the cycle nodes work, but the implementation has to be carefully done because of the possible problems superimposed on the superplot. On the other hand, 6LoWPAN does not require any synchronization between nodes, but their cycle should depend on the general e ectiveness of the load, and this makes it more di cult to plan the working cycle and therefore the power consumption. The results of experimental evaluations show that the 802.15.4 / ZigBee IEEE network is able to support smaller working cycles and provide minor nal delays with updating times that are a little closer to the theoretical value than the 6LoWPAN. On the other hand, the 6LoWPAN network shows that average delays are minor from end to end and thus, provides greater reliability; this is a lower percentage of packet loss.