Modeling the Energy Consumption - Part 1 - Introduction

As I told in my previous post, my target is finding an energy minimization scheme for transmission of information in Wireless Sensor Networks (WSN). In many typical WSN, the wireless nodes are separated by distances between 0 and 200 meters.

Considering Amplitude and Phase Modulations (PAM, QAM and PSK are included in this group), the energy spent per bit (E _b) needed to transmit information in a given bit error rate, is smaller for constellations with smaller number of symbols (M), or, equivalently, smaller number of bits per symbol (b), where b=log₂M. The drawback of using small constellations (ex: BPSK) is that, for a given available bandwidth, the time to transmit a fixed number of bits is longer than if one uses a constellation with a higher number of symbols (ex: 8-PSK).

The following graphs show those 2 characteristics. In the first graph, observe that for a fixed bit error rate (BER), let's say 10^-4, a constellation with smaller number of symbols (e.g. BPSK/QPSK) spends less energy per bit than a constellation with higher number of symbols (e.g. 16-PSK). On the other hand, observing the second graph, it is possible to see that the bandwidth efficiency (bits per second per hertz) of binary constellations is smaller than the ones of higher order constellations.

Figure 1 - taken from wikipedia

Figure 2

Considering a path-loss model, the received power is related to the transmitted power as expressed in the following equation:
Where P_Rx is the power of the signal in the input of the receiver, P_Tx is the one in the output of the transmitter, G_d = G₁d^kM_l is the power gain factor, with M_l the link margin for compensating imperfections in hardware and noise and G₁ is the gain factor at d = 1 meter [1].

So, the longer is the distance, the more power should be delivered to the transmitted signal.

Considering long distance wireless transmissions, like digital TV, mobile phones, satellites, the best strategy for saving energy may be to transmit using a small constellation size, because they require less energy per bit. The power spent by the RF circuit used to transmit this signal is much lower than the power of the signal itself and can be neglected in these cases.

On the other hand, for short distances, as it is the case of some WSN where nodes are separated by short distances, the power needed for the transmitted signal has the same order of magnitude of the power spent by the RF circuit used to transmit this signal, so, the power used by the RF circuit should be considered in these cases.

The tradeoff can be seen in the following way:

1. Use a constellation with low number of symbols and use a longer time to transmit information; or
2. Use a constellation with high number of symbols and use a shorter time to transmit information;

The first option minimizes the energy spent by the transmitted signal, but maximizes the one of the RF circuit, because it keeps the circuit turned on for a longer time. The second option maximizes the energy spent by the transmitted signal, but minimizes the one of the RF circuit, because it can transfer information faster and turn off the circuit after doing that.

Moreover, option numer 1 is the best choice for longer distances (observe the path-loss equation above), where signal energy is dominant, while option number 2 is preferable for shorter distances, where circuit energy consumption is higher than the signal one.

So, just making it clear: the objective of my research is finding a way to minimize the energy needed to transmit information in scenarios like the ones found in wireless sensor networks, considering both transmitted signal as well as RF circuit power consumptions.

This problem was studied in [1] and in many other previous papers. The issue is not solved and there are still many open problems.

If you want to know more details about the RF circuit power consumption model, please read [1]. I use the model developed in [1] to model the RF power consumption in my most recent three papers, that you can find here. In all my papers, there is an introduction, explaining that model.

That's it, that's all. If you have any questions or suggestions, please, leave me a comment.

More details in the next post, that I plan to write in the next 15 days.

Cheers,

BIBLIOGRAPHY
[1] S. Cui, A. Goldsmith, and A. Bahai, “Energy-constrained modulation optimization,” IEEE Transactions on Wireless Communications, vol. 4, no. 5, pp. 2349–2360, 2005.

My Research Theme

In this first post, I will make a brief introduction of what is my research here in Japan about. This first post is going to be useful specifically to make you understand the next posts and to put me closer to people that are interested in the same research topic. To structure my explanation, I will divide it in 3 main questions: WHAT, WHY and HOW.

1. WHAT?

In my master's course, I study ways of minimizing the energy used for transmitting information in an individual wireless link. I consider the scenario of energy-limited ad-hoc wireless networks, such as wireless sensor networks (WSN), where network nodes are typically separated from each others by distances between 0 and 200 meters.

2. WHY?

Minimization of energy is crucial for these types of networks, because they are constituted by battery-powered nodes.

If a wireless sensor network is used to monitor something inside your house, then, when one of the node's battery is over, you simply go to the supermarket, buy a new battery and replace it. Problem solved, done! Unfortunately, WSNs are more used to monitor remote and difficult to access areas. Battery replacement is impractical or impossible in this scenario.

In this case, much closer to practical applications than my house-monitoring example, the less energy is used to transmit information, the longer is the network lifetime.

Besides this reason, the present worries with the global climate changes demand the development of energy-efficient systems and that's what this research topic is all about.

3. HOW?

The focus of my research are individual liks. So, the investigation is done considering one single transmitter node transmitting information to one single receiver node. Those nodes are separated by a distance d, like in the picture shown above:



The distance d is a variable in my research, it ranges from 0 (zero) to 200 meters. My objective is finding, for a given distance d:
a) What is the minimum energy needed for transmitting one information bit?
b) How is that minimum energy reached?

• For a given modulation type (ex:M-QAM), what would be the optimum constellation size (M)?
• Considering that I am using an error correcting code, what would be the ideal rate to reach this energy minimization?
• What is the time needed to transmit all my information bits with minimum energy consumption?

---

That's a basic summary of my research. There are many details that I didn't talk here. They will be exactly the theme of the next posts. If you got interested by this topic, just get the pdf versions of my papers here. I recommend that you read the papers in the following order: VTC 2010, WBS 2009, GLOBECOM 2010, if you know what I mean. If you have no patience to read all, I recommend at least the VTC 2010 if you only wanna understand my research or the GLOBECOM 2010 if you already understood and wanna know my recent results.

That's it that's all. Next post, I will bring other details about my research.

Cheers,