物联网：人工智能是未来所有家电必备的

无论你称之为物联网(Internet of Things)或是机器对机器通讯(machine-to-machine communications)，还是它们的英文简写 IoT 与 M2M ，这种利用目前相对为主流的IP位置与无线通讯技术之应用，将会永远改变这个世界。 让许多──甚至是大多数──物体嵌入“云端” 功能，将改变这些物体的本质以及其衍生物；而且这些机器与机器之间的对话会是在暗中进行的，掺杂如手机窃听事件一般的、令人不安的回音。有些人可能认为 “人类好、机器不好”；或者是比较实际一点，认为“有些人类不好，但机器是中立的”；但无论如何这些都是简化过的想法，在做出结论之前我们要记得，M2M 这样的技术拥有为世界带来无限好处的潜力，只要能够适当地控制利用它。 飞思卡尔半 导体(Freescale Semiconductor)最近在英国苏格兰举行了一场研讨会，邀集甲骨文(Oracle)、IBM、Google等产业界跨领域高层，试图探讨如何创 造出有益的、可靠的且具安全性的 M2M / IoT 应用。飞思卡尔举例指出，每年全球因交通事故意外身亡的人数高达120万，但如果能利用车辆与车辆之间的通讯技术，就可以在驾驶人反应过来之前，更快地启 动车辆的煞车系统、或是更好地利用ABS，及时阻止碰撞事件的发生。 如此的考量以及其它潜在应用，导引出一个必然的结论：那就是 M2M 革命若真的发生，将会明确地以无线网络与安全性为发展基础。 除了说汽车内 的M2M技术将会实现，飞思卡尔的人还告诉我，目前市面上部分高阶车款内的线路总长达6公里(kilometers)，也就是说未来汽车内的子系统将朝向 “无线化”发展，才能减轻重量、节省能源。但车辆与车辆之间、车辆内部的无线通讯，再加上各种合法/不合法的干预、以及恶意软件等等问题，仍会带来有关安 全与可靠性的挑战。 此外，还得考虑IoT应用将如何登场；其中有很多恐怕只能利用现存的无线通讯基础建设，例如 Wi-Fi 、 3G 与 LTE，但IP位置本来就有限，越来越多对专用IoT频道的需求，有可能会严重排挤一般人与人之间的通讯。所以届时我们可能得问，到底有哪些应用是如此具 备经济上的吸引力，可以证明其需要专用无线基础设施的正当性？ 现在已经有一个利用M2M的计画，叫做“电子羊 (Electric Sheep)”；这是一种集体创作的抽象艺术，是安装在数以千计一般PC中的软件。当那些安装了软件的计算机进入休眠状态，“电子羊”程序就会相互进行通 讯，进行抽象动画的创作(编按：请参考http://electricsheep.org/)。 而除了安全性议题，我认为 IoT / M2M 技术的另一个问题，在于人类的信任与感觉。当车辆可以因为它知道“这么做比不做更好”，而自发性激活煞车系统，人类会有什么感觉？如果人们正想着改变一下 口味，但家里的冰箱已经按照一个重复性规则自动订购食材，心里又会是什么感觉？或者是当智能电网自动自发地关闭/开启家里的电器、只因为它认为现在该是睡 觉时间，我们又会有什么感觉？ 你可能喜欢或是不喜欢，但机器不会在乎这些。 编译：Judith Cheng 本文授权编译自EE Times，版权所有，谢绝转载 参考英文原文： Do machines dream of IoT?，by Peter Clarke

相关阅读：
• 十二五物联网最受宠，智能电网与医疗商机大
• 超低功耗ZigBee技术推动家庭和办公室物联网应用
• 物联网应用新商机：电视节目制作革命ZoWesmc

{pagination} Do machines dream of IoT? Peter Clarke Whether you want to call it the Internet of Things or machine-to-machine communications or shorten those buzz phrases to IoT or M2M, the application in this way of what are now the relatively mainstream technologies of IP addresses and wireless communication will change the world forever. The ability of many – or even most – objects to be embedded in the 'cloud' is going to change the nature of those objects and their derivatives. And the M2M dialogs will be done in a stealthy way, which many have disquieting echoes of the cell phone key tracking incident that is playing out as I write. It would be easy to oversimplify the situation to: humans good, machines bad. Or more realistically to: some humans bad, machines neutral. But it would still be a simplification and before we do it must be remembered that machine-to-machine communications have the potential to do almost unlimitless good in the world, if harnessed up appropriately. Freescale Semiconductor, recently held a one-day conference on the topic in East Kilbride, Scotland, attended by executives from Oracle, IBM, Google and others. It is trying to get close to the multidisciplinary teams that will be needed to create beneficial, reliable and secure M2M/IoT applications. For example, Freescale estimates there are 1.2 million road deaths a year around the world. Imagine the benefit of car-to-car communications that could apply brakes faster than drivers behind the steering wheel and apply them closer to the ABS optimum to shorten stopping distances and avoid collisions. Considerations of this and many other potential applications lead to the inescapable conclusion that an M2M revolution, if it is to come, will be based squarely on developments in wireless networks and security. Machine knows best Even within the car M2M will impact. The guys at Freescale inform me that the modern high-end automobile has 6 kilometers of wiring. The logic is that automotive subsystems must migrate to a wireless "wiring harness" to save weight and energy. But the co-existence of inter- and intra-car wireless communications, plus the potential for legal and illegal interventions and malware etc., will produce a security and reliability challenge. And it must also be considered how IoT applications will roll out. Many may be viable only if they can piggy back of pre-existing wireless communications infrastructure such as WiFi and 3G or LTE. But by their very nature the sheer volume of IP addresses could end up severely degrading human communications prompting the need for dedicated IoT channels. So then we must ask which are the applications that are so economically compelling that they can justify the construction of dedicated wireless infrastructure? One project that uses M2M already is called Electric Sheep. This is a collaborative abstract artwork in the form of software installed by thousands of people on ordinary PCs and Macs. When these computers "sleep", the Electric Sheep program allows the computers to communicate with each other to create and transphorm abstract animations. I described IoT/M2M security as a challenge, but beyond security there is another issue – human trust and feelings. How will humans feel about a vehicle that applies the brakes autonomously because it knows better than they do the consequences of not doing so? How will humans feel about a refrigerator that orders up food delivered on a repeat order just as they were thinking about a change? How will we feel about a smart grid that will selectively turn off - or turn on - appliances in our houses because it thinks we are sleeping? The humans won't like it and the machines won't care.