That is so so true for IoT. IoT needs expertise and usage in all fields, including arts science engineering sociology. (just a side note: our site www.cythings.com cover multi-disciplinary applications and news on IoT. Please feel free to check it if interested. Thanks.)
I'm pretty sure TCP/IP will be the chosen protocol. After all, TCP/IP is the foundation of Internet. The concerns of relatively large overhead can be reduced by various algorithm. I am pretty sure the increase of network bandwidth has already alleviated the overhead.
This is a great summary of the article. After 15 years of Internet, we have learned so many issues and are dealing with some of them. I'm sure there will be more issues as IoT really take off. Imagine a world of sensors everywhere, the potential issues/ concerns will be enormous.
Choosing the communications protocol is a major issue. It occurs to me that many embedded devices already identify themselves using JTAG. Maybe a combination of JTAG and low-power wireless would help. After all, the circuitry is already there.
We need to help set up multidisciplinary teams to harness IoT. I see this as the biggest challenge i.e. how to inspire medical doctors, biologists, economists, social entrepreneurs, not just engineers.